Oral History Transcript — Dr. Robert Frosch
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
See the catalog record for this interview and search for other interviews in our collection
Interview with Dr. Robert Frosch
Robert Frosch; July 23, 1981
ABSTRACT: Reviews Frosch's (b. May 22, 1928) education at Columbia University (PhD, 1952, theoretical physics) and, in detail, his varied career as a physicist and a science manager, beginning with his work as a research scientist at Hudson Laboratory (1951–3) and then as Asst. Director and Director of the Theoretical Division (1953–63). In 1963 he became Director of Nuclear Test Detection, Advance Research Project Agency, Office of the Secretary of Defense; in 1966 he was appointed Assistant Secretary of the Navy, Research and Development; from 1973 he served as Assistant Executive Director in the U.N. Environmental Programme; from 1975 he served as Associate Director of Woods Hole Oceanographic Institute; and from 1977 to 1980 he served as Administrator of NASA.
DeVorkin:Dr. Frosch, you indicated that you wanted to pick up another thread that had been missed.
Frosch:Yes, somehow or other, following down one line of discussion in education, I left out a thread that I think has some interest. My mother decided early on that we were to have music lessons. Now, why music lessons as opposed to art lessons isn't quite clear to me, particularly as I later concluded that I probably had some potential talent for art and none whatever for music. However, at a relatively early age, about the same time I started school, I began music lessons in the Conservatory for Progressive Music Education. It was a very good educational system. The school started small children making music before they learned anything. If you were so young you couldn't do anything else, you started with rhythm bands, and then you started playing the simplest possible instruments, simplified zithers or psalteries, etc., a very good system.
DeVorkin:The Kodaly system is like that.
Frosch:Is it? This really worked, because it was in effect a group play system that worked into group music making before you knew what you were doing. After that you had a necessity to learn notation. The idea was, now that you're doing it, wouldn't you like to be able to do it again? So let's figure out how to write it down so we can remember. That kind of thing.
Frosch:And it was really very neat, very good. After awhile, everybody started to learn piano, which I must say I didn't much like. I don't quite know why. Maybe it was practicing, which was more systematic, but there was something else. Playing the piano put too much machinery between me and the music. It is the only way I can describe it. I thought of it as a machine, and so I was working the machine, and the machine was making the music. And I really sort of broke with it. Somewhere in this — about the age of 11 or 12 — I developed a great desire to play the flute, which seem closer to the music. But what finished off the piano for me was the question of elbow attack. I don't know whether you play the piano, but there are all sorts of things about hand positions. One of the things was elbow attack. You bring the arm up and you flop it down and so on. I didn't understand what this was for, and I had a long argument with the piano teacher. The piano teacher didn't understand it, either, which is what made the break. I said, "Look, I know how a piano works. You know, you hit the key. What the key does is it throws the hammer, and the hammer hits the wire. I understand all of that. Please tell me the importance of where I hold my hand before it ever gets to the key. How does the piano know that?" I may be dressing it up a little bit, but that was the question. And the teacher didn't really know. The teacher only knew that it affected the sound, but not why. Now, the explanation is of course that it affects what you really do with your muscles just like follow-through. Follow-through doesn't affect the golf ball, but the fact that you're going to follow through does affect how you swing your arm. The teacher either didn't know that, or didn't explain it. I resigned in disgust. If they don't even understand the mechanics of how this machine works, how are they going to do anything? Anyway, that all got mixed up with the desire to play the flute. There was some discussion that I was too young. There was also the complication that I had extremely bad malocclusion and overbite, which was going to be corrected by orthodonture. In fact, the flute teacher from the conservatory took me on under the condition that I was to have it done.
Frosch:Yes, because it would have been quite impossible to form a proper embouchure. I learned to play the flute while having bad malocclusion and an overbite, playing the flute while having braces in my mouth. I got to be reasonably good and I enjoyed it. In fact, during the same year I took that postgraduate course in high school chemistry, I also went to Julliard and took six months or a year of advanced flute.
DeVorkin:Oh, that's marvelous.
Frosch:And all through high school and college I played piccolo in band and flute in orchestra.
DeVorkin:Did you continue to play in groups as you worked in Hudson and ARPA?
Frosch:Well, not really; I dropped out of practicing. I never became that much of a bug on it that I was driven. I became a listener. I practiced some and kept the skill up at least through the period at Hudson. It sort of dropped off when I went to Washington, just somehow came apart.
DeVorkin:That was your ARPA period?
Frosch:That was my ARPA period. I periodically will go back and practice enough so that I can play, but I don't play much. But I thought that was an interesting thread to recount.
DeVorkin:It's interesting in that you didn't want the machine to get in the middle.
Frosch:Yes, now that's extremely interesting. And I think that has some significance with regard to my attitudes and practices as a scientist and an engineer. In spite of the fact that I have dealt increasingly with very large-scale machinery operations, it still tends to be my attitude that the machinery is a necessity, but not something you'd really like to have. I have a certain passion for simplicity, at least, in the outlines of thought about a subject. I don't like the analytic attitudes that say, "We've programmed the computer and that's the way the answer comes out, or. "We did the analysis and that's the way it is. The outline can't produce the number or the detail, but the simple thought outline ought to be able to produce at least a plausibility argument that suggests that it's one way and it's not another.
DeVorkin:Yet you understood how the piano worked, but it couldn't be explained to you the nuances of body motion and why they were important.
Frosch:Well, they could have been, but I don't think that particular teacher really understood that it was a matter of body mechanics.
DeVorkin:I understand. Talking about feeling separated from the product, you mentioned in the first session that you didn't consider yourself really a manager at that time, or even now. Could you elaborate on that? First, for your years at Hudson?
Frosch:I've thought about that statement. In fact, I've thought about it recently, because I made one or two speeches that hinged on some administrative questions. And I guess I would describe it by saying more precisely that I consider myself a manager, but not an administrator. I have no personal interest in the mechanics by which the organization is run in detail. It is a machinery thing again. If I get interested in questions of accounting, it's because I'm interested in what the theoretical underpinings of the problem are. What is it you really mean by an accounting system in some abstract sense? What are the real constraints on an accounting system? I finally have translated it, where accounting is a kind of input-output transform with certain conservation properties. But other than that, everything accountants say is required is really free. You know, what's a budget? Well, a budget is a kind of transform on funding an input-output matrix. That's a rather crude version of what I have in mind.
DeVorkin:Do you think of these managerial functions in mathematical terms?
Frosch:Yes, but I want to describe both those ideas. One is this business of administration. I never sat down and systematically learned anything about personnel administration, organizational theory, etc., until I had been doing it long enough so that I felt that I had better find out what it is I am supposed to be doing.
DeVorkin:When was that?
Oh, probably some time around ARPA or thereafter. I remember distinctly when Jack Ruima was recruiting me. I ended up the discussion when he said he wanted me to come, by saying, "If you're looking for somebody who is going to administer an office, I'm not it. I can do that, but I have no interest whatever in worrying about all of that. My interest is only in what the place does and how to get people to hire somebody to do it for me." You see, I had grown up in Hudson Labs where, as I indicated, we had an excellent business manager who was experienced in dealing with scientists and engineers. I regarded that as a professional function, but not one I cared to perform, and one which philosophically I strongly feel is a support function. I really don't approve of organizations in which management is the function, and the substance is considered to be subordinate to it. That's what has happened to lots of U.S. industry.
I think that they have inverted themselves, and forgotten that they are in some particular business or other; they are just running a business, you know. I don't really think there is such a thing as an M.B.A. I just think it's kind of a funny fraud. There are certain administrative things to learn, but it is not a systematic subject. It can't be, because it has to be tied to what is to be done. The cash flow questions in an innovative place cannot conceivably be the same as they are in a place which is a distributor of previously manufactured items. Somehow that has gotten lost in modern theory. Now, the other thought. I was educated or indoctrinated as a theoretical physicist so that I am by instinct what would be called an analyst, although I happen to think that is precisely the wrong word. I would say I am a synthesist, rather than an analyst.
DeVorkin:It's interesting that the term analyst can be applied to a person, but a synthesist, there is none.
There is none, except in a very crude sense. I did tend to think of these things in rather abstract terms, and I think there is some virtue in doing that. In fact, I have some ambition to write a book on the kinematics of bureaucracy, by which I mean those things that happen in human organizations that are, so to speak, nonhuman. Things that are personality-independent, that have to do with structural, abstract and statistical reasons. I'll give you an example or two. I've written one theorem that interested me in the subject, and I think it was published in Spectrum, but I don't even have a copy of it. It's what might be called the staff theorem. The following observable phenomenon came to my attention. You have a smallish organization, let's say one individual with a staff, and everybody is overworked.
They are overstrained. And the conclusion is to add another staff person, and you do, and the system breaks down completely. It just collapses, becomes impossible. I've seen it a few times. I got to wondering — is there some underlying structure? And so, I devised the following problem: The easiest case is: I am a supervisor and I am supervising "n" people, who are in staff positions. Fairly important. How many ways can I supervise. I can close my office door. That is one way to do that. I can talk to them one at a time; that's "n" ways, or two at a time, and so on, on up to an all-hands conference. Now, if I make a simple assumption from my own experience, I do all of these things equally, so they have equal weights.
Then the question is, how much of this supervision activity is there as a function of the number of people. Now, just write down: There is one way to shut the office door: and "n" ways to talk to them one at a time, and two at a time, and so on, back to one way to have a staff conference. Add them all up, and you'll discover you the binominal expansion of one plus one to the "nth". In fact, the number of interactions with the staff is two to the "nth". Now, you know what happens when you add a staff man. You double the number of supervisory activities in one blow, because it's an exponential function, and obviously this causes collapse at a certain point. That also explains why the span of supervisory attention comes out around five people. You know, two, four, eight, sixteen, thirty-two. That's thirty-two activities; that you can manage. Six people is sixty-four activities. Ten people is one thousand and twelve activities. It has nothing to do with whether people are recalcitrant or anything. It's just a kind of combinatorial problem. There are perpetually people around who want to centralize things. The GSA is a marvelous example. Let's have one good organization that will buy everything for the U.S. Government. Then, of course, you discover it's a disaster. GSA is a kind of national disaster.
Part of the reason is in fact it's been a political dumping ground. But aside from that, people forget that there is what I call a combinatorial tax. That is to say, as the organization gets bigger, the overhead does decrease to a smaller percentage of the organization, but you have a new kind of overhead. You've got to coordinate the place, and that goes up in an exponential way. And those two curves cross over. There must be a natural limitation on how big an organization of a certain size can be, and still be increasing its efficiency, provided somebody really is managing it in detail. Well, that's a whole area of interest.
DeVorkin:When did you become interested in this way of thinking, this perspective on management?
Frosch:Well, I'm not sure. When I got into the Government situation in ARPA, I discovered that people were discussing systematically things that I hadn't thought about. I also discovered that my experience through the Hudson Labs gave me a lot of knowledge. I can give a mark point. I made a speech at Wright Patterson Air Force Base to a graduating class in management, within the first year or so of my becoming Assistant Secretary of the Navy. I was invited to talk to the graduating class of the Armed Forces Project Managers School at Wright Patterson AFB. And that was the first time that I formally had to stand up and make a speech of that kind. I hate to sit down and write. I remember distinctly what I did. Several days before, I simply wandered around the dining room and dictated a speech onto a dictating machine and had it typed. And that was an elaboration of everything I thought I had learned up to then about project management. I asked the question, why is management thought of entirely as a restrictive system, rather than one that makes it simpler to do things. I don't remember whether I connected it in the speech but I did personally relate it to why the Talmud gets ever more elaborate and restrictive.
Frosch:I have always thought of that because Hillel said, "You shall make a fence around the law." What happens is: the fence around the law in the "nth" generation becomes the law itself in the "nth plus first" generation. Therefore, each time a new fence is built around that law, after awhile the landscape consists entirely of fences. Of course, that's what happens in management. Every time an offense is committed, somebody makes a new regulation to prevent that offense. Nobody ever goes back and cleans up the system. After awhile, you have nothing but regulations. Well, that's another thread of thought.
DeVorkin:We have two articles around that time, one in I.E.E.E. Spectrum, September 1969, "A New Look at Systems Engineering." There were some contributions to that from members of my Navy staff at the time. It was basically mine, but added some things. Bill Raney did that, and Peter Waterman.
DeVorkin:Going back to the Hudson Labs times, how did you set long-range policy after you became director? Did these management structures enter into your way of thinking?
Frosch:We had "grown up" at Hudson; that's the only way to describe it. The structure was a reflection of the first directors and their experience, a kind of participative management. In fact, it was the natural way a group of physicists and engineers who respected each other would operate, particularly as the place was small. We decided what to do by sitting around and talking. That's the only way to describe it. Somebody would have an idea and wander into somebody else's office, and elaborate it until they either destroyed it or thought they had something. Then they'd try it out on somebody else. After awhile, suddenly out of the organization would come an idea. As director, I might do that, or the deputy director. Al Berman, somebody else might do that, and suddenly there was an idea. Now, beyond that there were certain themes that the laboratory had adopted out of this process. There was the theme of coherence — how nonstatistical is the ocean? The canonical question that came with that was how big a coherent receiver could you build for underwater sound at low frequencies? My specialty was low frequencies.
DeVorkin:Is this the same coherency as in coherent optics? Let me first change the tape.
Frosch:Let's put out a well-defined sound, a well defined wave form somewhere in the ocean, and make the following inquiry: Let's take two receivers, and start them out very close together. They will receive the same sound, which will be the signal, plus some noise. As you begin to separate them, a new question arises. Is the ocean along the two paths by which the sound came from the source to the receiver identical or nonidentical, particularly with regard to its properties of change with time? Aside from noise, let's assume a very high signal-to-noise ratio. Is what arrives at the two hydrophones identical or similar, and how similar? This of course determines how big an antenna it pays to build. If you get a separation where the signal is quite different between the two, then you might just as well collect energy and add up the energy. You shouldn't be collecting a wave form, because you can't get that coherent addition. It is like coherent or incoherent optics. You see, I grew up in underwater sound where all signals were born coherent. Now, people who grew up in optics, at least not in the most recent optics, grew up in a world where everything was incoherent. That led to some confusion in conversation.
Frosch:On underwater sound at low frequency, all band widths are infinite, as far as people who work with optics are concerned. We were quite happily working in the band from 10 cycles to 300 cycles. That's a very large band width. So this coherent question was one important theme, and the nature of the noise was another important theme, for the very simple reason that these are absolute fundamentals. If the ocean is coherent, you can do certain things with the signal; and if the noise has peculiar properties, then you can do special things with signal-to-noise. This was the general intellectual program that Gene Booth outlined: if you find out what nature is doing, you can probably find out what out what you can do. This philosophy led us into some elaborate experiments. The last session we ended by talking about the coherency, the first experiments.
DeVorkin:You said you were going to begin talking about it.
Frosch:Well, we've led our way back to it. There was a debate going on with regard to building passive receiving systems. I'm going to leave out some numbers, because I don't know whether they are classified or not, but it won't matter for the discussion.
DeVorkin:Sure. Fair enough.
Frosch:In building operational systems, the question was, how long did it pay to build an array? In fact, there was a doctrine that said it didn't pay to build it very long, because the ocean is incoherent. The time change property along different paths is such that it is incoherent. We were taking the opposite position. At frequencies of tens to hundreds of cycles we could not see why that should be the case. We thought they were being confused by the multipath properties, many rays, and interferences among the rays. So we set out to produce a crusher of a demonstration. That's the only way to describe it. Not only were we going to do an experiment. We were going to do an experiment on such a scale that nobody could argue any more about whether we were talking about tens of feet, hundreds of feet or miles. We were just going to do it on a scale of several thousand miles, period!
DeVorkin:Was the entire lab working on this? Was Gene Booth working on this, directing this for you?
Frosch:No. this was much later when I was director. Those heavily involved were myself, Alan Berman, who was the deputy, and C.S. Clay, who is now I think at the University of Minnesota.You can look him up in the Journal of the Acoustical Society of America. He is publishing in acoustics. I don't remember whether Ivan Tolstoy was involved or not. I think not.
DeVorkin:I was going to ask you about him later.
Yes. Well, the idea for the experiment was very simple. We brought up a guy who was considerably older from the physics department at Columbia, Dana P. Mitchell. Mitchell had concluded earlier in these discussions that underwater sound needed a very precise single frequency source, in order to do controlled experiments. Remember, these were the days when you didn't have computers and you barely had digital electronics. You had some tube digital stuff, but crystals were not something that you bought in a $20 watch. They came in ovens, and so on. The control of high-power acoustics was not in good shape. He did a very ingenious thing. He took a device that was called the A Mark 6B sound source, which was a mine sweeper source, a horrendous object. It had a large steel box containing either a five or a ten-horse-power D.C. motor and a crank shaft, and two pistons. The pistons were about a meter in diameter each. They were set in big rubber gaskets in the side of the box with the crank shaft, so that they simply moved in and out in opposition, changing the volume of the thing. The acoustic output wasn't good. The output was best around 30 cycles, so we habitually worked in 30 to 100 cycles. This was the standard device, with a D.C. power supply.
By changing the voltage, you changed the speed of the motor, and hence the frequency. Dana put an extension on the shaft of the D.C. motor. On the shaft he put a one-horse-power A.C. synchronous motor, ran the A.C. for the synchronous motor out of a dual mercury thyrotron power supply, which was run by a countdown from a good quartz crystalcontrolled oscillator. You started wherever the quartz crystal was. It was probably 50 kilohertz, 100 kilohertz, and you had a good counter-scaler, a regular nuclear instrumentation kind of thing. Then the countdown. That controlled the thyrotrons, which fired and were smoothed through a filter that produced the A.C. So then you'd run the D.C. motor up to the speed you wanted. Using a lissajous pattern between the output of the D.C. motor and the output of the thyrotrons, you could get to the point where the pattern stood still, and then lock in the A.C. motor. That produced enough power to control the wobble of the DC. Most of the power came from the D.C. motor, and the control came from the A.C. motor.
DeVorkin:It was an enormous synchronization.
Frosch:Oh, enormous synchronization, a very clever device, very simple. It was ridiculous from the point of view of current control systems, but it was beautiful. It was solid. You could take it to sea, and so on. That gave us an oscillator with an effective Q of 9,000. In the acoustics business at that time those frequencies were phenomenal.
DeVorkin:That's very accurate.
Frosch:Q of 9,000 means that the variation is such that you go 9,000 cycles before you vary as much as one cycle. In this instance, this is at 30 cycles. That means you're going for 300ths — you've got a source which is stable to the tune of 300 seconds before you're out by one cycle. You're going for minutes before you have to worry about it. In fact, it was very good. The idea of this experiment was to determine if the ocean breaks up low frequency sound. We proposed to make a very narrow band sound, receive it at several places, then find out if we still have the wave form.
But in order to make sure that everybody got the point, we wanted an array, that is, to add coherently the output of the hydrophones from different places, as well as receive it. So we employed the long lines department of AT&T, and we picked four existing Navy hydrophones and connected them through leased telephone lines into the lab. There was one in the Bahamas, one in Canada, and two in the U.S. We hooked them all into Hudson Labs. Then we sent the ship out with the source at a precise frequency and sat happily in the laboratory for two weeks, 24 hours a day, with a duplicate of the crystal on the ship. Besides, we could send what the crystal on the ship was doing by radio, so we had all of that data. We were able to demonstrate happily that what you saw as the variation of the sound frequency was the effect of the Doppler shift of the ship's velocity. The Doppler shift was just enough to put it off frequency.
Occasionally the thing would glitch and then come back in. But for hours and hours, you would watch the slow phase change. Just by looking at that spacing, you could tell the velocity of the ship. And in fact, you could take the output of the two hydrophones, whatever the distance was, and put them together and they would behave perfectly coherently. This was good enough so that we could tell the ship's velocity. And it got so precise that at one point we thought we knew when a particular helmsman came back on the wheel, because his control wasn't as good, and the Doppler would wander. I think we verified that. We had a lot of fun with it. We came back saying, you want a demonstration that the ocean stands still, at least as far as 30 cycles? Boy, does it stand still.
DeVorkin:Was this the Navy you were saying this to?
Frosch:Well, it was partly the Navy, but it was really the community of underwater sound people who were interested in this kind of problem.
DeVorkin:Who was particularly skeptical?
Frosch:I think the Bell Labs people were skeptical. In a sense, we won the argument. It did not affect the systems as much as it should have, but it became quite important for some classified reasons later. All I can say is that fruits of that have subsequently been demonstrated and used in Navy systems.
DeVorkin:The papers that identify this activity began around 1957, the same period as Project Artemis. Was this part of Project Artemis?
Frosch:This was done several years before that. It wasn't really.
Frosch:Yes, I think the work was done earlier. I don't really remember. There was a sequence of experiments.
DeVorkin:Berman, Clay, you and H.B. Sherry, "Preliminary Report on the Correlation of CW Signals and Large Hydrophone Separations."
Frosch:Yes, that must be right. I had remembered it as earlier, but it could well be that date. It was relevant to Artemis. I don't think it was, strictly speaking, part of the Artemis contract.
DeVorkin:Shall we move on to Artemis, then?
Frosch:I don't know whether you've got the name Medea in there, but I ought to note the name for a variety of reasons. The Navy wanted an acoustic survey of the Norwegian Sea. It was called Project Medea, and it was not separately funded.
DeVorkin:Not separately funded?
Frosch:Meaning, up until Artemis, everything that Hudson Labs did was a single task in a single contract. It was just block-funded. You didn't go to the Navy and say: "We've got this experiment; will you fund it?" We argued our program and our money for the year, and ran the laboratory. We justified what we were doing on the basis of what we had done. That had great virtue. It meant that, if in the middle of things, we decided that the line we were taking didn't make any sense, we could do something else. We just went and did it, and then told the Navy guys. If they screamed and jumped up and down, we'd have that argument, but we were in control of what science and engineering was to be done. It would be on our head, you know. If we screwed up, then we wouldn't get any more money. It was very straightforward.
DeVorkin:Did that ever happen?
Frosch:No, no. There was an annual argument about overhead rates with the university, and we always asked for more money than we got. But on the whole, it was all right. So I suddenly found myself, by agreement, the chief scientist for an expedition to find out about underwater sound in the Norwegian Sea. This is in what is called the GIUK Gap, the area of the United Kingdom, Faroes, Iceland, Greenland, and north from there into the Great Norwegian Sea Basin. This basin, by the way, had never been completely mapped. No one knew what the bottom shape was, and no one at that point knew much about the underwater sound properties, transmission or noise.
DeVorkin:That was an awfully crucial area for submarines, too.
Frosch:Yes, that's why it's kind of interesting. Anyway, I was asked to put together the expedition. There were some Hudson Labs people involved. I guess we used the USN Rehoboth and an EPCR from what was then called the Underwater Sound Lab.
Frosch:It's a designator. It's a small electronic support ship. It was a Navy ship. And there was a cable layer the Bell Labs had under charter. I don't remember the name of that one. No, we used the Navy oceanographic ship USNS Rehoboth.
DeVorkin:Which class is that?
Frosch:It's a small seaplane tender, an AVP. It was called an AVP in World War II, which was really a seaplane personnel ship, a 200-foot 2500-ton. There were five ships in the expedition. There was a DE that was being used. We went up to do a rather complex thing. We had to lay hydrophones and measure sound transmission, and do some high frequency work. There were five ships milling around.
DeVorkin:They were all American ships?
Frosch:All U.S. ships, yes, a mixture of Navy, Coast Guard and private ships. I think there was an MSTS ship in there somewhere. That stands for military sea transportation service. It was really a peculiar, mixed bag.
DeVorkin:Did Great Britain know about this?
Oh yes, all the Allies, but this was a U.S. show, essentially. We put into port in Iceland, Canada, the UK, and Greenock, Scotland. I found myself playing two roles. One was chief scientist, which really meant to get people together and plan what had to be done technically so that we got the right collection of data. The Navy was to supply an operations officer who was going to worry about how to move the ships in response to all of this, the whole logistics thing. Shortly before we were to set sail a month before, the Navy couldn't produce the ops officer. I don't know whether the guy got sick, or what, but I ended up being the ops officer, at least for the purpose of planning. And I suddenly found myself being told that the Navy wants to know exactly what you're going to do with these ships all summer, and they want to know on Monday. And this is Thursday or Friday.
By then I knew enough about the ship business, having been running the Hudson Lab ships. So I just sat down with the chart and took the science program and made a two-month op schedule. It was kind of a pain; it was very tedious, but I did it. I had to coordinate the ship's capacity and fueling requirements and so forth. We put together a detailed operations calendar. The calendar had five columns going down two months; it listed the tasks, the distances, the steaming times, and the allowances for weather and so on. It was the ops plan. And we went off and spent the summer wandering around the Norwegian Sea. It was interesting. It was a fun kind of operation. There was one very funny tale.
The Navy decided that this project was classified. Initially, only messages dealing with the substance of the science were going to be classified. That was all right. Rehoboth is the name of the flagship. I think it's named after Rehoboth, Delaware, but I'm not sure. That's the AVP; that was the flagship. Anyway, at the last minute they decided everything was classified, including the logistics messages. Nobody was to know where we were and that was difficult. The Rehoboth was a small ship, and it didn't have any encryption machine, any ability to code messages. But they made sure that two of the young officers on the Rehoboth were cleared and qualified to put things into code by hand from the books, where you look it up and do it letter by letter. So that was a big grunt and groan. Now the funny event was that the flagship had to collect the necessary information from the other ships by blinking lights at appropriate times, in order to send to the Navy in Scotland a request for supplies upon landing at Greenock. It's the port at the west end of the Clyde Estuary, the real port before you go into Glasgow. And so, this message goes out asking for 500 pounds of four-way beef. Four-way beef is a standard Navy packaging scheme in which there are standard proportions for steak, hamburger, roast, stew meat, etc. So the message goes out, but apparently something went wrong in the encryption. Back comes a message about 24 hours later, which duly decoded says: "Re your request for fifty thousand pounds of beef; if correct, please explain." We suddenly conjured up visions that all over the British Isles, cattle were being rounded up. But we had that kind of amusement. And a lot of very bad weather.
DeVorkin:Yes, I can imagine.
Frosch:There are a whole bunch of sea stories. But we collected the data, and I had to sit down and write a report. It wasn't science at all, or even engineering. It was a kind of crude survey, but it produced data that the Navy later used for some purposes. It was in the "Department of Odd Jobs." And in a way, it was interesting that that there was a community of research places. The Navy could come to us and ask us to design a survey and do it, and nobody had the slightest difficulty in putting together a university contractor, a Navy laboratory, a commercial contract at Bell Labs. I think there were some others involved. It was good practice for Artemis. As chief scientist, I had to preside over this group of people from different labs that I really didn't have control over except by their courtesy and the agreements. I had to get them to agree that we're going to do this experiment and not that experiment, and this one is contingent. There was a certain amount of fighting at sea. I wanted to carry out certain things of higher priority than the Underwater Sound Lab guys wanted. We had a little bit of quarreling by message over whether they could go do their experiment, or the agreed experiment, and so on.
DeVorkin:What year was this?
Frosch:This was the summer of '55.
DeVorkin:You were not yet the director of the lab?
Frosch:I was not yet director. I was assistant director or associate director, or whatever it was.
DeVorkin:And this was when the lab was still a block grant lab.
Frosch:Yes, it remained that way. When Artemis came in, they decided that this project was separate. We wanted to keep the general accounting separate, so they made it a separate task in the same contract. Actually, the fact of task is more important than contract, because the contract was really an omnibus contract with the university. They made a general agreement with the university. and then all of these things were subtasks, so they only had to negotiate overhead.
DeVorkin:I see. Well, let's talk about Artemis. Would you give me a description of the purpose and goals?
Frosch:Yes. I think I'd better go through a little historically to make the relationship between Artemis and the previous experiments clear. There was a commercial contractor, it might have been the predecessor of Ford Aerospace, led by a retired Naval officer. They had an idea that they claimed was going to lead to a very long-range active sonar system. Here I'm going to be vague, because I simply don't know what the status of this stuff is. I don't think it will affect what you want to know. This project was based on a principle that made perfectly good physical sense, because it didn't violate anything. But we looked at it and concluded that the numbers were hopelessly wrong. The report had been optimistic in certain estimates and you could not, in fact, build the system the way he wanted to build it.
DeVorkin:This proposal was made to the Navy?
Frosch:It was made to the Navy, and then the Navy asked us to review it. We worked in the role, among other things, of consultants. Since we were on contract, there wasn't any fuss about it. We looked at it. In fact, we convened meetings. What was then the Committee on Undersea Warfare of the National Academy convened discussions of it. The general conclusion was that it won't work. Because the subject came up and had to be reviewed, it got people very interested in building a very large, very long-range active submarine-detecting sonar. And so, we went off and thought about that. Again, when I say we thought about it, we thought about it individually. We talked to each other, and we consulted the community. The community by then was an organized thing. There was a group called the Undersea Warfare Research and Development Planning Council, which was Navy sponsored. It was kind of a club, a classified conferencing club of those labs that were dealing with the undersea warfare business. It focused on the acoustic detection of submarines, active, passive, any other way.
DeVorkin:Who organized that?
Frosch:ONR, I think. I was a member of it. It was very pleasant, very productive. We argued problems, devised experiments and traded data. But also, very pleasant, a clubby group of lab directors in the same business, who would meet three times a year and thrash out their business. We also gave advice to the Navy, some of it gratuitous, some of it taken, some of it not taken.
DeVorkin:This was more a policy kind of a group?
Frosch:No, it was more a group that worried about the science and engineering. For example, somebody would report that an experiment ought to be done. I'm thinking of a particular one, and again, I don't know the clasification status. I know it turned into a system. I remember saying, "What do you want to do an experiment for? We have, in effect, done that experiment fifty times. We know it will work. It is obviously the right way to build a system, but there are severe engineering difficulties of such and such." They wanted to do the experiment because you will never convince anybody, just because it's science. So they made some submarine detections, and we all applauded and said. "Well, we told you it would work." And they said. "Yeah, but we know it will work now!" They were right, because the Navy then picked it up, and in spite of tremendous engineering difficulties, it eventually has become a recognized system principle.
DeVorkin:This is not Artemis?
Frosch:This is not Artemis. Let's get back to Artemis.
DeVorkin:You were talking about the undersea warfare.
Frosch:All right. I don't remember whether it was done formally or not, but Hudson Labs, with agreement from others, began work on a sonar. We believed there was a good possibility that we could build a system with a good detection range. It was to be very large in an engineering sense, and rather expensive, but there was a chance that it would have a significant detection range. Now, there was a professor of physics at Harvard, Frederick V. Hunt. Ted Hunt, who was one of the great old hands of the underwater sound business from World War II and thereafter. Anybody in the acoustics underwater sound business remembers Ted Hunt. He had been following what we were doing, and he said that the motto of this submarine searching with acoustics ought to be "an ocean an hour." Now, "an ocean an hour" for the simple reason that sound travels in seawater about a mile per second. So in 3600 seconds, you're going to go 3600 miles. You sit in the middle of the Atlantic, you can go 1500 miles out and 1500 miles back, so an ocean an hour. He was also the one who first described a million square miles as a "Navy acre." That was the scale in which we were thinking. So, we finally came in with this proposition that we think there is something to be built. We designed an experiment and had a set of ideas about how to do it. By then, I was director of Hudson. I don't remember exactly how I got to be elected or appointed as the leader of all this, probably because I have a big mouth. Jim Wakelin was the Assistant Secretary of the Navy, R and D, and so I explained all of this to ASNR&D, who bought it.
DeVorkin:Was that the first time you went and discussed things personally with someone at that level?
That was the first time I had gotten to an assistant secretary. I had talked to chiefs of Naval Research and admirals, and so on. That was a little bit rarefied. I knew there was such a creature. but that was about all. I wasn't quite clear on what he did. I guess I knew Jim Wakelin, because he had been around the underwater sound business before that, but didn't know him well. Essentially, I was sent in to make a pitch. We had decided this was worth doing, and ONR thought it was worth doing. We had an estimate of what it would cost, which was high by the scale of anything we were dealing with. It was more than double Hudson Labs per year. In fact, that's roughly what it did, was double us, but we were only doing a piece of it.
There were lots of others. I don't remember what the scale of money was, but it must have been five or ten million dollars a year, which is a lot of money for underwater sound in those years. We were running Hudson at a couple of million a year, plus a ship. The Navy provided two ships. So Navy decided to do this. I found myself with a new budget in another task, and a new authority, and another group of participants. Again, I was chairing a group of participants from other laboratories, and trying to decide how to build this giant object. It's a long, complicated tale. We went through a lot of agony with the sound source. That was the first attempt to build it out of gigantic magneto strictive scrolls. Some magnetic materials have the property that they contract in a magnetic field — you just throw the magnet field to it and it changes its properties and contracts and expands. You know, it's the magnetic equivalent of piezo-electricity. In fact, it's known earlier in underwater sound than piezo electricity, because magneto striction was an early principle for building sound sources. And this was a job that Lockheed in Burbank was going to do.
DeVorkin:Were you subcontracting?
Frosch:Yes, we were subcontracting. We were doing all sorts of things. There were cases in which we were not subcontracting, but the Navy was contracting directly, and we were the technical officers. We were also subcontracting, handling money and writing contracts. We had to get an additional guy just to handle our contracting.
DeVorkin:That's what I was going to ask you. You must have expanded staff at that point.
Frosch:We did. We doubled the lab.
DeVorkin:Now, is this double the lab, with what kinds of people?
Frosch:Scientists and engineers, a little bit with administrators, but not much. Well, we doubled the administration. Instead of having two people, we had four. You know, it was that scale. That's not fair; there were more people than that, but it was that kind of thing. Most of the increase was in engineers and scientists, principally engineers. I was going to give you the scale of the sound source. In order to hang it we got the Navy to take a 10,000 ton T2 tanker out of mothballs and cut a sea chest down through it. The receiver involved a major sequence of cable-laying operations. It was the most complicated thing that had ever been put in the ocean at the time. It was very, very large; it had hundreds of hydrophones on it, thousands. It was quite a manufacturing and engineering design operation.
DeVorkin:These were standard hydrophones? There was no special design for them?
Frosch:No, they were built out of a Hudson Labs design. We demystified a lot of the hydrophone design business. The first low frequency hydrophones were designed by people who were used to high frequency. They assumed that problems at high frequency would be problems at low frequency. It immediately became obvious that that wasn't true. At high frequency, it is hard to make a hydrophone that is not comparable to the size of the wavelength, so you have all sorts of effects. At low frequency, it's hard to make a hydrophone which is as big as one percent of the wavelength. So there's nothing that can happen to the hydrophone at low frequency, except being squeezed, and that simplifies life greatly. But at first it wasn't recognized, and we looked at it and said, we don't understand why this won't work.
Frosch:We had a lot of interesting experiences. There was the time we received a shipment of 3,000 ceramic cylinders for hydrophones, and discovered that they were fine, except for one thing. None of them exhibited any piezo-electric properties whatever. That was kind of an interesting detective game. We discovered that by buying in large quantities, we were affecting the production capability of the people who made barium titanate cylinders. There wasn't much demand, so there wasn't much capability. Nobody had planned the logistics. It reflected all the way back to the manufacturing capabilities of Titanium Corporation of America, for which this was a small side business. In order to pump out as much powder as required, they had simply said, "Well, nobody will know the difference whether it's this mesh or that mesh, so we'll omit the final grinding." And that turned out to be crucial.
DeVorkin:Wasn't that something they could be reprimanded for?
Frosch:Oh yes, but it was a minor incident. We didn't. It's just a funny incident and we dealt with it very easily Obviously, they accepted responsibility and there wasn't any problem with it. It's interesting because I had been having that kind of contact with the engineering and the industrial problems and the logistics. That sort of thing had happened to me long before I got to ARPA or the Navy. It bears on this question of how you learn management — you learn it the hard way. You learned it by discovering that if you smelled something funny when a contract was bid, it paid to call in the auditors. They wouldn't give you a very good answer, but they could find out if it was fraudulent, or if there was something funny, or if the guy had a bad record. I found that sort of thing very useful. Not that I knew much about it myself, but again, I had people that I could depend on. I could simply go talk to them. You know, I could go talk to Marion Johnson or Frank Mongelli, who was the Artemis business manager. And Frank had been through that kind of thing before, and we'd sit down and put our feet up, and he'd after awhile say, "Well, you know, I think there may be something fishy here; let me find a way to find out.
Frosch:"Bless you, my son, go find out." And he would do that. We had a very easy working relationship. Yes, things are getting complicated. We're in a pretty large scale engineering operation: ships, ship modifications, surveys, cable layers. We had to buy a Texas tower for a cable termination at Bermuda.
DeVorkin:Was that one of the Glomar Challenger type towers, or are you talking about something else?
Frosch:Yes, but just a tower. It was that vintage of offshore oil tower. It was just when the J. Ray MacDermott Company was beginning to plant these things offshore and drill. This was '55. We were going to put our receivers on the slope of the reef southeast of Bermuda. We were using everything we knew about underwater sound. The source was going to be hung from the T2 tanker, the "Mission Capistrano," where the swallows come back to. We had to run cable from the place we had chosen for the receiver to Bermuda. This was in the days when you had to put repeaters in the cable for our purposes and we needed a place to put the amplifiers. For a variety of reasons, we needed an amplifier house, and we ended up deciding the most economic thing was to buy this Texas tower and have it put on the reef. We called it Argus Island. I guess the reef was called Argus, I think.
DeVorkin:There was no connection with Project Argus?
No. So we put this Texas tower up. That was all done by contract, of course. It was an interesting example in field logistics. What you really needed was one person there for the care and feeding of the electronics. Of course, you can't have one person alone. So you have two people. But you're going to operate around the clock, and by that time you've got enough people, so you've got to worry about the food. Then you've got enough people, so that you've got to worry about emergency services. And then you've got enough people so you've got to worry about making beds. Anyway, you end up with close to a dozen people (laughs).
It was an interesting lesson. The termination place and the real laboratory was going to be on Tudor Hill in Bermuda. The Navy, which owns some property under the treaty, gave us an abandoned World War II gun bunker under the top of Tudor Hill. This was maybe 150 feet above the ocean. And the Navy noted, by the way, that the heads in this gunbunker flushed on fresh water. Fresh water is always at a premium in Bermuda. It's caught in catchments. In fact, there was no problem, because the hill was its own catchment. In principle, somebody in Washington said, "There's a shortage of fresh water on Bermuda — part of the responsibility of the laboratory is to convert these heads to salt water flush." Hank Beck was our chief scientist at the Navy Oceanographic Office in Bay St. Louis, Mississippi. Hank went down and looked at it, and came back and said, "Bob, this is ridiculous. It is going to cost us an awful lot of money, because we've got to put in pumps, Orangeburg pipe up the hill, and so on. It's going to cost twenty, thirty, forty thousand dollars for nothing, because the whole top of the bunker is the catchment. We're not going to run out of water." Back to the Navy: "No, you've got to do it." Hank and I kicked it around. He says, "I've got an idea." And he went away and came back with a computation which we duly submitted to the Navy, after which they finally threw up their hands and said, all right.
Hank made a computation which proved we could flush the heads with gin at Bermuda prices for ten years before we'd amortize the Orangeburg pipe. In those days you bought gin in Bermuda for eighty cents a quart or something. He demonstrated it was cheaper to do it with gin, and we put it in as a proposal. The Navy just said, "Fine, fine; forget it. Let's not discuss it any further." That gives you some of the flavor. You know, we were having a very good time, and working very hard, and having very, very difficult problems. My introduction to engineering at Hudson Labs came because you couldn't go and buy equipment off the shelf. You had to go and build it. So we found ourselves engaged in building the stuff. Well, that was the Artemis thing, and we began to lay stuff on the bottom. We had to invent a whole system for laying things that had never been designed before. We wanted to build a three-dimensional array.
Frosch:That is, hydrophones in three dimensions. For a variety of reasons, we had to split out arrivals and do a whole bunch of things. There wasn't any very good way of laying such a thing. Harry Sonnemann had been our chief electronics guy, and he was a key guy in a lot of this. He was on my staff at ARPA, and in the Navy and in NASA. He is still at NASA.
DeVorkin:So he moved with you?
Frosch:Yes, we didn't move together, but the circumstances were such that he came around. You know, I would need him and so he moved with me. He took on a lot of the design and bossing the contractors who were building these hydrophone arrays. In fact, he largely thought up the ideas for how to build arrays that were collapsible until you put them on the bottom, and then they erected themselves and became a string of three-dimensional structures. People were explaining to us all along that this couldn't be done, and that they had proved five years ago you couldn't do this. And step by step, we were doing it. Now, I actually left before we got to the stage of demonstrating any of it, so whatever demonstrations, and there were some successful ones, were done after I left.
DeVorkin:So the time of Artemis is really from the late 50s all the way through '63 when you left?
Frosch:Oh. it was still going on. I don't think they really finished up until '65 or '66.
DeVorkin:Is that system in use now?
Frosch:It was never capitalized on. I think there were some successful experiments, but people looked at the cost and what could be done. For a variety of reasons, some of them policy reasons, it was decided that they wanted the knowledge in the bank, but they weren't going to build any such systems.
Frosch:Several things have to be said about that. One is that if we were going to design such a thing today, it would be very different. The signal processing was a terrible problem for us, because there was no computation capability that was up to it. We were taking the signals from hundreds of hydrophones, and wanted to compute on them and combine them, and do coherent processing, using things we knew about the source and so on. All of this would be done quite happily by a modest computer today. But we had terrible problems doing anything.
DeVorkin:Did you use any computers?
Frosch:We were looking at optical computers at that time, rather special purpose things. You see, pushing a light pattern through a lens is taking its Fourier transform. Once you have done that, you can manipulate things in Fourier transform space and then invert them. There was a whole system of doing this, which I never understood very well. But there were a couple of guys in the project, particularly Ross Williams, who were working on that. And where that ended up, I don't know, because I lost track of it after I left.
DeVorkin:This was actually a kind of matched filtering technique?
Frosch:It was an optical matched filtering technique.
DeVorkin:Using laser lights?
Frosch:No. There were no lasers yet.
DeVorkin:So optical match filtering and noncoherent light?
Frosch:Yes. By putting the signals on film, and then using the film as coherent filters, even though the light was incoherent, you were essentially using a coherent filter on it. It involved using the incoherent light as a combining thing.
DeVorkin:But you didn't really try to use digital computers?
Frosch:Well, we were in the digital computer business, but there was nothing that was of a speed and scale that you could use for this kind of signal process.
DeVorkin:The 709 at that time wasn't fast enough?
Frosch:If you look at the memory and the computations required, they weren't anywhere near it. You would have to digitize the acoustic signals on the order of 102 or 103 sources, and then perform bit by bit computations on them, including lag times. That's a lot, and is now comfortably done by a reasonable sized parallel processor. But people were sort of mumbling parallel processing, but they didn't have any hardware. You couldn't do it with a batch machine. What we had, I guess, was a GE225.
DeVorkin:I don't even know that machine.
Frosch:Well, it was going to be hot stuff, but it never got to be. It was a perfectly good machine. We were using it for scientific computation, not much for signal processing.
DeVorkin:So you weren't too heavily into a dedicated computer shop at this time?
Frosch:We were not a big computer shop, never were a big computer shop. We used it. The stuff was there.
DeVorkin:Yes, at Hudson.
Frosch:We started out with a CPC, and then moved to the GE225.
Frosch:The IBM card programmed calculator.
DeVorkin:Yes. Now, Artemis then took a major portion of Hudson Lab efforts through the rest of your tenure.
Frosch:Well, half of it. It was roughly half and half. It was an interesting social situation. There were a group of people at Hudson who didn't want to have anything to do with Artemis. They felt that taking on this project meant that the success or failure of the single project in its own terms would mean destruction of the lab, one way or another. If it were successful. it would destroy the basic lab, because everything would flow into it. And if it were unsuccessful, it would destroy the lab, because the entire lab would be marked off by the Navy. That's not what happened. In the end, when I was ASNR and D, I destroyed Hudson Labs, because of what Columbia's attitude towards it had become.
DeVorkin:Yes, I'd like to move into that sort of discussion now.
Frosch:Okay, but it's much later. This is much later.
DeVorkin:I know. That was already the late '60's. Yes, but let me first ask you where the word "Artemis" comes from.
Frosch:Oh yes, I was going to tell you. Yes, I told you about Ted Hunt. Since Ted Hunt was sort of the father figure, in a sense, of this whole business of doing long-range, large scale sonar, the first name that we chose was Project Diana, because she was the Roman goddess of the hunt. It turned out there already was a Project Diana. It was the project of getting a radar reflection from the moon. Diana was the Roman name, so we went back and chose the Greek name, which was Artemis, who was the goddess of the hunt. So that's the in-joke on Project Artemis.
Frosch:But there was a certain trend to that. Now, let me see if I can get the names right. If I don't get the names right, then there won't be any joke. There was a kind of sonobuoy.
Frosch:Sonar Buoy is what it came from. it was a passive buoy; it just listened. The guys at Bell Labs invented a complex system, the details of which don't matter, which they called Julie. They called it Julie, because there was at that time a cafe singer in New York named Julie something or other. And the point about Julie was that it made passive buoys active.
Frosch:Yes, exactly. There was a certain amount of this sort of thing going. So, using Artemis as the goddess of the hunt was a fairly natural in-joke.
DeVorkin:Well, in 1961 there was considerable New York Times coverage of Project Artemis.
Frosch:Yes, I guess there was. I had forgotten that.
DeVorkin:This was also the Cuban crisis period. What reaction did you have to that New York Times coverage at that time? Without readingit myself, I don't know exactly what they were saying. Do you recall the whole situation?
Frosch:Well, rather vaguely. I don't even recall whether they had it accurate. I don't recall where they got the story. I think we'll have to go to the morgue and review what it says. I remember the articles. I remember the main reaction was faint embarrassed annoyance, because they were spreading on the newspapers stuff that we had all been rushing around very carefully to conceal the work at the Navy's request.
DeVorkin:Do you remember how they got the story?
Frosch:You know, there is no way that you can conceal the existence of a 10,000-ton tanker that has funny superstructure. There's just no way to do it. I mean, you could sneak around in the dead of night, but it isn't going to do you any good. (laughs)
DeVorkin:Where was it being altered?
Frosch:That's a good question. It seems to me it was altered in Philadelphia or New Jersey some place, and where were we basing it? It must have been based by MSTS somewhere in Brooklyn. You know, it was some place where, if you'd drive along the right expressway, you'd see it. It may have been in the Brooklyn Navy Yard, or some place. Somebody stumbled on it and got curious, and built up a story out of it. We'll have to go look. It's just vague. I have a vague image of some pictures and so on.
DeVorkin:Yes. I'll hold that one out. Who was Ivan Tolstoy?
Frosch:Ivan Tolstoy is a theoretical physicist and acoustician, whom we hired at Hudson Labs. I don't remember where he turned up from. He'd been a graduate student of Maurice Biot's who was a mathematical physicist. Tolstoy's specialty was computations and complicated boundary condition situations in what I'll call acoustics. But he had been doing this work somewhere in the oil exploration industry. He had worked for Exxon, or one of the firms. He decided he wanted a more academic kind of atmosphere, but didn't want to teach, and didn't like the industrial thing. We were sort of made to order, a research place, with no teaching involved. And we wanted that kind of theoretical capability around. Ivan came and did that for quite a while, essentially on his own as an individual scientist. He took on the task of elaborating the theory of acoustics in layered media with funny boundary conditions. That's really the problem, and he's been at this all his professional life. I still see an occasional paper of his. He's apparently living some place in Scotland and turning out theoretical work, and I guess he must be consulting.
DeVorkin:I see. I was just curious.
Frosch:I haven't had contact with him for awhile. He is a very interesting guy, just as a person. The name does stem from the Tolstoy family. He was born, however, from a branch of the family that had fled to Germany during the Russian revolution, and had its own set of curious adventures during World War II, very funny stories. He was in Paris, and he was a White Russian with a French passport, German parentage, and clearly no good was going to come of this in Germany, so he walked from Paris to the Swiss Border, walked across the Swiss Border, and said, "Here I am. Intern me." And they looked at his passport and threw him back over the border. Anyway, through some contacts of his father's he ended up in some curious way with the Swiss agreeing to accept him as a Russian. They put him in an internment camp, but as a Russian soldier. Again, through some manipulation, they bureaucratically decreed him an officer, which meant he could go on parole and go to the University of Bern, which he did. Part of the parole was the requirement that he come to town every week to the police station and register that he was still there, and get his chocolate ration card. And he goes down to the thing, and the guy says, you're not wearing the flag of your nation in your lapel. It turned out, he was the only one who was decreed a Russian officer. And Ivan says, "Well, you know, you understand my peculiar circumstances. I don't really want to wear a hammer and sickle. Besides, the Communist Party is not legal in Switzerland." The guy says, "All I know is that it's in the rule book that, if you don't come with the flag in your lapel, I can't give you a chocolate ration card and renew your parole." So Ivan goes out and his landlady served him up one and he appears at the police station with the flag. And the guy says, "My god, don't you know the Communist Party is outlawed in Switzerland?" So anyway, there got to be a big climbing up through the Swiss bureaucracy. Finally, Ivan claims. somebody made a Swiss bureaucratic decision: Ivan would wear two flags. On the front, he would wear a French flag, and on the back he would wear a hammer and sickle.
DeVorkin:That's a marvelous story.
Frosch:Anyway, it's a crazy story, it's a marvelous story. We had some interesting people. We had a guy named Roberto Frasetto who came to us as a diver, but turned into a scientist and engineer. He's an Italian, U.S. citizen. He had been one of the first officers and organizers of the Italian UDT's. He was captured in the raid on Malta when his buddy was blown up. A very interesting guy. The last I knew of him, he was running the Instituti Di Grande Massi, which is the branch of the Italian NSF, or the Academy of Sciences, which had the responsibility for saving Venice. But I don't know if he is still there.
DeVorkin:Let me ask you about the relationship with Columbia University while you were there.
Frosch:It was vague.
DeVorkin:Yes. How were the organizational relationships between ONR, Columbia and Hudson? I knew they were a bit vague, but were they smooth?
Yes, they were basically smooth. The Navy had a contract with Columbia University, which covered a variety of research. At one point it included the Nevis Laboratories, or pieces of it, and Lamont and us. We were all tasks under this contract. It was administered by a contracts and grants guy down at Columbia. In fact, we wrote the proposals. We did all the contacts with the Navy. We did all the work. Columbia handled the business, the formal accounting. All our accounting went downtown, and they did it, and they did the payroll We had the use of the libraries, and so forth. There was a continuing struggle to try to get into some closer relationship with the university. We had a feeling that it would be good to have direct contact with the university. We always wanted some kind of status in the university.
One issue for awhile was TIAA pensions. That did get resolved so that we got into TIAA, but we never got any kind of adjunct professor status or anything like that. People sort of wanted that kind of contact — not tenure. That was obviously ridiculous. None of the departments quite wanted to get involved. There were people in the physics department who were mildly interested, but nobody spent much time on it. So we were pretty much on our own. There was also an annual contention over the overhead. In particular, the Navy auditors were always bothered by the laboratory overhead. Auditors have funny ideas. Somewhere I have a letter from an auditor from ONR, which says in effect: don't buy a book next year. You bought a book last year. You already have a book. That was the flavor of it. The idea was that since there's a Columbia Library, obviously, you don't have to buy any books at Hudson Labs. That was kind of silly. The Columbia Library was important to us, but it was twenty miles away, and it didn't have anything decent in the way of an underwater sound collection. Anyway, you need reference books.
DeVorkin:What was your overhead?
Frosch:Twenty-one percent or twenty percent was the overhead.
DeVorkin:Was that high at the time?
Frosch:No. It was a funny overhead, because it didn't include the building, which was a separate rental lease, and didn't include our materials, didn't include our light, heat, hot water. Really, it was just the overhead for what Columbia did, which was accounting, payroll, library, certain legal responsibilities and so on.
DeVorkin:Any computer services?
Frosch:Only insofar as they may have the machines to process payrolls and so on. It was low. Overhead was then running 100%. You know, I don't know whether it was G&A overhead: I never bothered to understand that set of distinctions. For a commercial contract, it would have been 100%; for a contract on campus, it probably would have been 80%.
DeVorkin:So it was a low one.
Frosch:It was a low overhead. It was really very amicable. Every year there would be an argument about whether the 3% for the library was allowable, and so on and so forth.
DeVorkin:Yes. It does sound vaguely similar to the JPL-Cal Tech situation with NASA.
Frosch:Yes, I think it probably is, except my impression is that Caltech has always paid more attention to JPL. They've had the same sort of argument. Should JPL people have faculty relation, should the faculty people work in the JPL, etc. it was always kind of unsatisfactory, but somewhat more attention from Caltech than from Columbia.
DeVorkin:Did you get any attention from the physics department?
Frosch:No, no. They vaguely knew we were there. I think there was a vague feeling that some people were afraid we would embarrass them. I had my own personal contacts. I talked with Charlie Townes or I talked with Henry Foley, but I couldn't talk with them much, because they weren't cleared. I'd bring them in an occasional physics problem which we'd talk about or something, but not much.
DeVorkin:You mentioned the word embarrassment there. Was that because that you were doing military work?
Frosch:No, not that it was military. There was a faint feeling of bother about the classification, that the campus shouldn't be involved in classified stuff. It was all right, you know, if we were out by ourselves, and a vague feeling that somehow or other classified stuff couldn't quite be respectable. The people who knew us in the faculty knew that if we were doing it, it probably made sense. And there were some people who were vaguely proud of the fact that we were doing something that might be said to be useful. Gene Booth's comment when I was trying to understand the relationship when I first got there: he said, "Well, the first thing you have to understand is that Columbia is owned and operated by poets. If you have that clearly in mind, a lot of things will be more clear to you." Actually, it wasn't. It was run by historians, but it came to much the same thing. Now, I'll give you the capper. The capper was when I had decided to go to ARPA. It was clear that I had to resign and make sure that proper arrangements were made for selection of a director and so on. The first thing I discovered was, no one had the faintest idea who I was supposed to resign to, because they didn't know who was supposed to be supervising me. The only one we ever talked to, other than my personal friends, was the guy who ran contracts and grants, and he was clearly not the supervisor. So our business manager and the contracts and grants guy went off and did a research project to find out who it was I was supposed to resign to. They finally produced John Krout, who was the vice president for administration or something. And I went and resigned, and he had to have it explained to him.
DeVorkin:Oh. amazing. That's amazing.
Frosch:What's really interesting is that, at that time, just from the point of view of money handled, Hudson Labs was the second largest fiscal entity in Columbia University, exceeded only by the medical school.
Frosch:But you know, it was a contract and grant thing, so nobody was paying any attention to it. As long as we didn't bring any disrepute on the university. they were proud of us. But curiously enough, they exercised only what I'd call business oversight. It never occurred to anybody in the university that they ought to exercise a kind of professional oversight. Nobody came around and said, "We're going to have a visiting committee for Hudson Labs."
DeVorkin:There was nothing like that?
Frosch:Well, I'm trying to remember. There was some discussion of visiting committees, but I think we generated our own.
DeVorkin:Through the Navy?
Frosch:No, we just went out and got some people, but I really don't remember.
DeVorkin:Do you remember who the people were?
Frosch:No. I don't remember anything about it, except that there was a discussion for a while of visiting committees.
DeVorkin:So there was no Navy oversight committee, or anything of that sort?
Frosch:No, I don't remember any formal body that used to come around and say, "We're going to look into what you're up to." You have to remember that the contact with the Navy was really very good. There were some ONR contracts people in New York. Steve Feraris was in Washington. That was for business arrangements. They worried about auditing from the Navy point of view, and there were the usual limits. You'd want to buy an instrument that costs more than $10,000 — they'd worry about that. They kept the property record. But the spirit of the thing was that we were part of the ONR community. There were guys in ONR, Naval officers and civilians, who felt responsible for us. And I could be on the phone with them. I'd whip down to Washington to tell them what I was up to. They didn't have to worry about whether they knew whether something good or bad happened. We were telling them. We were writing letters. They knew all about what was going on. In a sense, there was this community of people who were looking over each other's shoulders. and publishing in a classified way. Nobody bothered with the idea that there should be a formal visiting committee. That's rather curious.
DeVorkin:Was there any involvement with the Cuban crisis embargo, blockade?
Frosch:I wasn't there then. When was that?
DeVorkin:I thought that was '61.
Frosch:Oh. I guess I must have been there then. No. That was day-to-day military operations. We didn't get involved in that. If it had heated up to something, then we might have. The last crisis situation that I was involved in just before I left Hudson to go to ARPA was the search for the Thresher.
DeVorkin:I was going to ask you that.
Frosch:And that's fascinating. It illustrates a number of things. It illustrates the nature of this community of people and their relationship to the Navy. I can tell you precisely where I was when I heard about the Thresher. We were having a meeting of the Undersea Warfare Research Development Planning Council in Washington, I think, at the Naval Ordnance Lab. We were at dinner, you know, we always have a dinner. We had a number of Navy officer guests, senior officers, vice admiral kind of thing. And we knew these guys. And, was I chairman? I don't even remember; maybe I was chairman. Somebody came in and called the vice admiral out. Somebody else came in and called another admiral out. Admirals were popping in and out. At the end of the evening, after the dinner sort of formally broke up, one of the admirals told us what the problem was. By the next morning we, the committee, had outlined the search for the Thresher, divided up the immediate tasks, and worked out with the Navy what all the labs were going to do. And Hudson ended up with a curious responsibility. We ended up running the communications for the Thresher search, because the Navy did not have adequate equipment, particularly, to deal with civilian ships. So we used our radio room and our single side band capability. That was when single side band was not so common, so the first job was to get enough single side band equipment to equip all the research ships. And we began to operate. I went out. We used the Mission Capistrano on that. We used the Gibbs. Everybody used all their ships. I was out at sea on the Gibbs on that search for two weeks just before I left and went to ARPA.
DeVorkin:How did the structure of the search go? I mean, who was in on creating the basic structure? Can you tell me what techniques were used?
Frosch:Everything anybody had, is the easiest way to describe it. There were all kinds of experimental and search sonars. The principal one was the side-looking sonars that had been developed out of the mine sweep business. They are now common. They were kind of exotic then. In fact, we had one built especially for us by Westinghouse, which was building the side-looking mine sweeper sonars then. That, I think, was what finally found it, not ours, but somebody's. No, it was a photograph that found it. Woods Hole was simply towing camera sleds.
Frosch:There were terrible navigation problems. There was no good navigation off of there. We were using for navigation, as I recall, a loran A-line and a loran C-line, because you couldn't get across other loranCs, because one of the stations was too far away. We were using everything there was, and it was in a terrible, curious mixture. One of the problems was that, even if you knew where you were, we did not have very good techniques then for knowing where anything we were towing was. You might be towing it on a mile of wire to get it close to the bottom. And you sort of knew where it was, but when you made a turn, it went all over the place, so there was a lot of reconstruction to do. Then there was a funny thing. The Navy decided that, even though it was mostly civilian ships and so on, they really couldn't stand seeing it done the way were doing it, where it wasn't quite clear who was in charge. We were saying, "Well, I'll do this and you do that." So they appointed a Naval officer in charge, Frank Andrews, a very good guy. He had a doctorate in acoustics from Catholic University, but he started to act like a Naval officer in charge. While we all knew him, nobody was going to put up with that. We kind of detoured around him slightly until he got the idea that the way to operate with this was with us. You couldn't go around telling these guys what they were supposed to do.
DeVorkin:Was there any competition for getting to the Thresher first?
Frosch:Everybody wanted to be the one that found it, but it was not so much competitive as "how fast can I do this, and how fast can I do that." People just sailed their ships when they were ready. It was not that kind of a competition.
DeVorkin:It wasn't a hostile competition.
Frosch:Oh no, no; a very cooperative kind of competition. We divided things up and people were working on each other's equipment. We were providing lab space and house room for guys from the West Coast so they could ship gear in and come into our lab and deal with it. We'd truck it and take it on our ship. We ran other people's gear from the Gibbs, a very cooperative community. You know, people would yell at each other and argue about what the right thing to do was, but then they'd settle down and go do it. I may be coloring it rosy in recollection, but I don't really remember any big problems with people. We were really fighting the problem and fighting our own frustration. We were sure if we could only get this piece of equipment to work consistently, it would do the job, but we kept having to haul it up to fix it, because it was lab gear.
DeVorkin:I'd like to ask you two more questions about Hudson, and we can move on to ARPA. First, what caused you to leave Hudson Labs? Second, could we jump ahead and talk about the demise of the Hudson Labs before we talk about anything else?
Frosch:Okay. I didn't leave Hudson, I joined ARPA. I mean I had no intention of leaving Hudson. I got a phone call from Jack Ruina, who was the outgoing director of ARPA. He's at MIT. He had been director of ARPA and was about to leave to go back to MIT. He needed a director for nuclear test detection and said to Bob Sproul, who is now the president of the University of Rochester, "Before you come in. I will have recruited a director for you for nuclear test detection." which was important because the Limited Test Ban Treaty was just about to be signed. I got a call saying, "There's this job, and we want you to take it. We think you're just the right guy. Would you at least come and talk about it?" And I said that I had no intention of leaving Hudson. I finally said, "Okay, I will come and talk about it. And I told my wife Jess, "Well, we've talked about it a lot of times. You know I don't want to live in Washington. We're not going to Washington. But this sounds important enough that I've got to go refuse it in person."
DeVorkin:You said you were married again by this time?
Frosch:Oh yes, I was remarried in December of '57, and at this point we had two small children. One was born in '60 and one was born in '63. We had lived in Brooklyn, and we bought a house in Nanuet in 1961, so we were all set. I went down in June or July, and it was very hot, I remember. Jack Ruina sort of talked me into it. As I may have mentioned, I ended up saying, "Look, if you want somebody to administer an office, I don't know anything about that. But if you want somebody to manage your program, that I can do." Anyway, he talked me into it, and I came home and said, "Well, I'm afraid I accepted this job."
DeVorkin:How did he talk you into it?
Frosch:He just made it very exciting and interesting, and a new kind of thing to do, and terribly important. He just talked me into it. I don't remember it in detail, except I remember starting out saying, "I came to say no." And he said, "Now, let's walk around and talk about it." So. I'm saying, "Well, okay, I guess." The importance was a piece of it. There was a good deal of violin playing about national policy and the importance of nuclear test detection, and the treaty, and so on and so forth. It involved a certain amount of flattery. I just remember starting out by saying, "no." and ending up saying, "All right, I'm committed. I'll do it. When do I have to come?" And coming home and saying, "Sorry about that, but I'm afraid I accepted the job."
DeVorkin:Were you saying this to your family?
Frosch:To my wife.
DeVorkin:What about the people at Hudson, and Columbia?
Frosch:Well, I didn't even think about Columbia, but I thought a good deal about Hudson. Part of the thing with Hudson was that Allen Berman, who had been a graduate school classmate of mine, had come on some years before. We were close colleagues and worked very well together. When I became director, I asked him to become my deputy director. We had run these long-range experiments together, literally. He built the equipment and both of us ran it. It was that kind of lab, the director took a watch on the equipment and went to sea, and so on.
DeVorkin:Yes, but he was the equipment builder?
Frosch:He was an experimentalist. He was a student of Polycarp Kusch, a magnetic beams guy and an experimentalist, and a very good one. It was obvious that the place was not going to be headless. Allen could run it as well or better than I could. What was in my mind was that Al Berman would take over and find his own deputy director. In fact, that's what happened. He later went on to become what's called chief scientist, technically, of the Naval Research Laboratory. That's where he is now. If you want to talk to him about Hudson Labs, he's handy too.
DeVorkin:Oh, I see, yes. All right.
Frosch:I had a feeling that I was leaving a big part of my life, but I wasn't leaving it without anything. There was an obvious way to continue. Besides, it gave Al a chance to move up. I was probably more unhappy about the unfinished state of Artemis than anything else.
DeVorkin:Was it moving on smoothly?
Frosch:Oh. it was moving on as smoothly as it ever did, with its problems and so forth. It was not in a crisis or anything like that. But I always come in the middle of the movie now, and leave in the middle.
DeVorkin:Oh. I see. Let me ask then, just to close out on Hudson Labs, what caused you to decide, when you were Assistant Secretary for the Navy Research and Development, to close Hudson?
Frosch:All right. I can't reconstruct all the details, but roughly, the following happened. Al Berman was selected to become director of NRL. Then he left. The university then was really faced with the responsibility to find a director. First, they got a guy named Jim Heirtzler, who was at Lamont; subsequently, he was at Woods Hole. Jim was a first class research geophysicist. He came over and was terribly uncomfortable with the fact that there was classified work there. There may well have been at that stage more classified work than needed to be. He wanted to change a lot of things. I don't know exactly what happened while Jim was director, but I got very uncomfortable reports from the people who were there. He didn't like the spirit of the place. Now, there had always been a rivalry between Lamont and Hudson, but I don't think that was it. It was kind of a friendly, occasionally vituperative rivalry, but that was not of the essence of it. Nobody was out to get the place or anything like that. I think he viewed the laboratory very differently than we had. He wanted it to be much more of a basic geoscience laboratory than we had thought of it. We had thought of it as doing the science of the ocean, so that you could do the acoustics and the underwater sound. And if you did good science, and could publish some, that was dandy. But you really had an eye on an application out at the end. There's a philosophy here that maybe we ought to come back to sometime. The question is, what is the relationship between basic and applied research? And the first thing to be said is: there is no difference between the two.
DeVorkin:Who said this?
Frosch:I say this. There is no difference between the two in terms of what the work is. The difference is in the attitude and the motivation. That is to say, basic and applied are not the same as pure and impure. You've got some of that in some of my speeches.
There's a motivational dimension, and there's a purpose of work dimension. My philosophy is that you start to do applied work. You are basically using the science that you know and applying it, but if it is work with any complexity and difficulty, you're going to keep stumbling up against pieces of science that are missing. And so, you're going to have to go back and at least work in particular areas of basic research. There is no sense in calling them applied, because you just don't know everything and you've got to expand a basic piece of knowledge. Then you do that in the ordinary spirit of basic research, and maybe you solve the applied problem. And also, inaybe it leads you off into some interesting basic research directions that you had never noticed until it was motivated.
It's like the development of analytical mathematics in the 19th century coming out of problems like ripples on water and the calculus of variations coming out of problems on the shape of constrained cable. It's that kind of thing. You know, it may become pure mathematics, but the problem started because somebody had an applied question. I get very upset about people who tell you that you shouldn't do applied work, that you really ought to be doing pure research. You're neglecting a source of problems. He wanted a much more pure kind of thing, and that led to some problems. There was also some funny opportunity for money from the University of the State of New York. At that time it was barely a university; the regents controlled education in New York. Somebody decided that New York ought to go into oceanography. Various campuses were competing to do that, and Columbia somehow got the idea that they could use Hudson Labs as the vehicle by which they could grasp onto this money and build up an oceanography program. I found myself faced with a situation where something was happening to Hudson Labs. It was serious enough that the ONR people, the Navy people, came to the Assistant Secretary. Whether they would have come to the Assistant Secretary if he hadn't been the ex-director of Hudson Labs, I don't know. They said, "We've got a problem. Columbia wants to take this thing and march off in this direction. Will you resolve it? You know the Columbia people." and so on.
I found myself in a situation where, reluctantly, I concluded that Columbia no longer wished to have Hudson operate as a vehicle which would be useful to the Navy. It might be useful to the Navy in a sense of being an important national asset in oceanography, but that was not why the Navy was putting block funding into Hudson Labs. I went up and talked with Grayson Kirk. It was either Grayson Kirk or Bill McGill. I think it must have been Grayson Kirk, or if it was Bill McGill, it was Bill McGill two milliseconds after he took over from Grayson Kirk; I don't know. The correspondence would tell, but I don't know where it is. Unfortunately, anything that I thought of as an official record, I never decreed to be a personal record, and I left them all where they were. The only files I ever took away are travel bills, or something that I might have to account to the IRS for, or personal correspondence.
DeVorkin:Right. Did this have anything to do with the Mansfield Amendment at that time?
Frosch:No. No. it was all prior to the Mansfield Amendment, although it's faintly reminiscent of some of that. Mansfield then is another interesting tale.
DeVorkin:That was about '67, though, wasn't it?
Frosch:No, it was later. It seems to me it was later than that, but the two are separate in my mind.
DeVorkin:Okay, but it would be nice if we could talk about the Mansfield Amendment later.
Frosch:That's interesting, yes. Yes, I'd like to talk about that, because it is instructive in a policy way.
Frosch:The conclusion was that I couldn't see any way to save the place, and still preserve what I thought was my honorable and proper task as Assistant Secretary. What was worse was I felt that, if I acceded to what Columbia wanted to do, I would be destroying the place anyway. I'd be leaving them with a name and a few of the people, and a reputation, but they would not be doing what the spirit of the place was. I reluctantly concluded that we had to terminate that relationship. The main pain of it was that I still had a number of friends at Hudson, and I was shooting their base out from under them. There are still a few of them in the same building; they built up a small independent firm that I lost contact with.
DeVorkin:Was there any contact with Berman on this? He had already left?
Frosch:He had already left, and he and I talked about it. He had come to the same conclusion that Columbia was going to do something with it that did not make any sense.
DeVorkin:The person who replaced Berman had this idea that there shouldn't be any classified activity going on, and that was the feeling generally on the Columbia campus, Columbia being a very activist area at that time. Was there anything in that?
Frosch:Yes, it was clear that there was something being driven by the campus activism about no classified research. As I remember, there were some of the usual weird articles in "Spectator," which is the Columbia student newspaper, speculating in very wild ways about what it was that went on at the classified place with the armed guard. There was always an armed guard. I was always worried about the armed guards because they were usually retired local policemen, very nice guys, but I was not sure that I liked the idea of their having pistols and bullets.
DeVorkin:This is how you felt?
Frosch:That was how I felt, but we had to have them, and they did watch over the safes and all of that. The place had a reasonably good record for keeping its classified stuff.
DeVorkin:We'll come back for a while to Hudson Labs when we retrieve the New York Times coverage. We'll give that a crack, to get the New York Times coverage.
Frosch:Yes. Let me say that I discovered there are lots of people in the Navy who are still feeling nostalgic about Hudson Labs. Last night I talked to the Sea Grant Association, and I ran into Dick Stevens, who used to be at ONR in New York. He remembered Phil Shandler, that's the head of the office of ONR in New York. He remembered all of those relationships fondly. You know, it was pleasant.
DeVorkin:Yes. Now you moved out to ARPA.
DeVorkin:And I'd be interested first, as a transition, in what ways were the funding mechanisms similar in ONR and ARPA? You drew your money directly from DOD?
Frosch:Yes. They were similar in the sense that it was one source funding. We were responsible to the Director of Defense Research and Engineering, Harold Brown first, and then Johnny Foster, although really it was always Harold Brown when I was in ARPA, because the transition to Johnny came just before I went into the Navy. But it was not block funding. You didn't get the funding for ARPA in one lump, you had to put up a budget and a program, and you did the the pieces one by one. And, of course, we had to go to Congress.
Frosch:Well, It was a piece of the appropriation, and it was defended to the committees as ARPA. The director of ARPA went. Now, when I was office director, I would tag along sometimes and sit in the back. When I became deputy director, then I went up with Charles Herzfeld who was the director, and participated in the defense, although I did relatively little of it, because Charlie is a good talker, articulate, and he did most of it. But I was there, and I got a taste of testimony, and so on.
DeVorkin:What was the organizational structure of ARPA like? You were director of nuclear test detection?
Director for, I guess is the way it is, which is nowadays a common term. The structure was that there was a director and a deputy director. The director was an executive level-5. The deputy director, I guess, was a PL-313. Do you know about PL-313's? Well, at one point, in order to simplify getting scientists and engineers, particularly, into Civil Service, there was a Public Law 313. It produced the ability to hire scientists and engineers, specialists, categories, in an undefined grade between 16 and 18, defined by the salary, the salary not to exceed GS-18. That was what was required to get me there. It wasn't all that much in those days; it was under 25 grand, as I recall. So there was the director and the deputy director. Then there were four directors and an assistant director. I think that was Bill Godel. He functioned in a way that was never quite clearly defined, and it all ended badly, if you know anything about that.
He ended up in jail. It's an interesting thing to be faintly involved in. He was doing certain special projects. He functioned as assistant director, and I was responsible to him for some things, but mostly not. I mostly dealt with the director and the deputy director, Bob Sproul and Charles Herzfeld. If he called and said, "Would you do this?" I would do it, if it was a reasonable request. He never made an unreasonable one to me. What happened was that he had been heavily involved in the counterinsurgency, particularly in Thailand. He had been involved in that and had lived there. I don't know the details, but somebody in Thailand gave him a house. There was a court case and he was convicted of fraud, or fraudulent use of government property. It happened quite early on when I was there. I rather liked him and respected him, but there was something funny and it just went sour. In fact, I know when I found out, because Bob Sproul called me at six in the morning to tell me. I was in a motel in Cocoa Beach, because I had gone down to see the launch of the first Vela satellite, which was mine because I was the director for nuclear test detection.
DeVorkin:That's certainly something I want to get to.
Frosch:I had inherited it just before launch so I was involved in it. Okay, now I was responsible for my office. I had to make a program, make a budget, fight it through the director, and defend it to anybody else. It was clear what my charter was. We now had a limited test ban treaty; it had been signed that summer, and therefore, I was the responsible agent, although there was an Air Force responsibility, and considerable battling over who was responsible for what. Before, I was the responsible agent, although there was an Air Force responsibility, and considerable battling over who was responsible for what. As far as everybody was concerned, I was responsible through Bob Sproul and Charlie Herzfeld to Harold Brown and McNamara for seeing that the U.S. had the technical capability to know whether anybody exploded anything in space or under the ocean. I think my work under the ocean drew their attention to me. Now, detecting explosions was easy. That's trivial.
DeVorkin:It's a rather large event.
Frosch:There's a very large noise, and your main problem is reverberation. We have lots of data from the Bikini underwater shot which was sufficiently loud that it was used by several people very cleverly to verify the position of all the sea mounts in the Pacific Ocean by multiple echoing off things. So I knew the Navy system cold. I knew what we had to do, which was almost nothing. It almost amounted to just putting recorders on certain things to be sure that we weren't going to miss anything, and by god, we didn't miss anything. That was easy. The space thing I knew nothing about, but I had to learn the whole Vela business. I concluded that was going to work all right, if we got the satellites up.
DeVorkin:Who was the specialist in your office on the Vela satellite?
Frosch:The guy who was responsible was a young Army officer by the name of Hillman Dickinson, who was then a Lt. Colonel, and is now a Lieutenant General, and is a very good guy. It's worth talking about Hill a moment. He was an Army officer stationed at West Point. He was assigned as an instructor at West Point.
Frosch:Physics, I think but I don't remember. While he was at West Point he managed to get permission to split his time. So he commuted from West Point to Hoboken, and got a doctorate in physics at Stevens; a very smart man.
Frosch:Pretty good. And I guess Hill was the guy who was responsible for the Vela satellites.
DeVorkin:For the conception of them?
Frosch:No, no, that had been done by other people. Remember ARPA was the original NASA. ARPA was really set up to be the military version of NASA before there was a decision as to who was going to do space.
Frosch:I came in well after that, but some of this stuff had lingered on. That's why the Vela thing had been at ARPA. That's how nuclear testing detection really got into ARPA, I think, because of Vela. They suddenly discovered that was well in hand, and they now were responsible for seismology and underwater sound. There was nobody in the office that knew anything about that, which is how they got to me. You could ask Jack Ruina, but I presume that's right.
DeVorkin:There was still a little bit of Argus going on, just as you came in.
Frosch:Argus, I remember the name.
DeVorkin:Nuclear blasts in the atmosphere.
Frosch:Oh yes, but that was by then an Air Force responsiblity.
DeVorkin:So you had no contact with it.
Frosch:Oh, I had contact with it. But the whole question of airblast waves and barometric measurement was being done over in the Air Force, and it was in perfectly satisfactory shape.
DeVorkin:I'm interested in the Vela satellite system. The first launch occurred just as you came. I don't know: we would have to look up the dates. It couldn't have been just as I came, but it was within the first year, maybe within the first six months. I came in September, and it must have been in the spring or something. I don't remember.
Frosch:My recollection is that I learned the Vela system, what it was about, and so on, but in a limited way. It was all done, and my problem was only, did I believe it. Did I think we knew how to use it. It was going to become operational, and I didn't fuss over it too much. I certainly didn't fuss over the launch thing, because it was clear that was contracted to the Air Force and they were going to do it. I couldn't affect it, you know; I was a guest at the launch. I was not even in the block house. I was over iri the industrial area, which is all perfectly reasonable.
DeVorkin:You had no decisions to make about the types of detectors?
Frosch:No, no, the thing was in the last stages of being put together.
DeVorkin:Then Hillman Dickinson would be the person to talk to.
Frosch:Hill would be. Yes, he certainly would be one to talk to. He may remember it slightly differently, but he would be a good guy to talk to about that.
DeVorkin:Later on in the series of Vela launches, wasn't there some time given over to astronomical observations?
Frosch:Oh yes, but that had been planned early on. They were particularly looking at the gamma ray bursts, in fact, for optical and gamma ray coincidences, timing and so on. I remember being one of those, whether it was my idea or not, who realized that the timing was good enough so that we could come back with arrays again. We could use the two Vela satellites as a pair of timing detectors. They were far enough apart so that you could get a direction for an extrasolar system gamma ray burst. Some of that was done. There were people interested in doing it and collecting the data. During my period there, there was the first detection of an extragalactic gamma ray burst, and a lot of excitement. I was very amused, because I had been the apostle of very large arrays in underwater sound. Here I was running the largest possible arrays I could have worked on.
DeVorkin:Were the scientists who were interested in collecting this data in ARPA?
Frosch:No. no. It was probably Ricardo Gianconni, or George Field, or somebody at Harvard. It seems to me it was the Smithsonian Astrophysical Observatory group, but that can be verified. It is interesting that I recall very little of this, and I think the reason is very simply that I put relatively little attention on it. It was obvious that Hill Dickinson and some others were very competent, and understood it completely. It was going smoothly, and I had other problems to take care of. So I just didn't worry.
DeVorkin:Was there any difficulty making the scientists aware of the nature of the detectors on Vela, so they could understand what kind of observations they were perceiving?
Frosch:That had all been done before I came on board. And I think, to some extent, the instrumentation may have come out of those very scientists.
DeVorkin:That's very interesting. The scientists interested in astronomical applications provided the detectors.
Frosch:Yes, because they were the people who were worried about gamma ray detection in space. Hill Dickinson is probably a better source for that piece of history than I am.
DeVorkin:Okay, where is he now?
Frosch:You could also ask Jack Ruina.
DeVorkin:Is he at ARPA?
Frosch:He's at MIT. He was the director during the period that ARPA was built up. Hill Dickinson was in the Washington area up until around last September or October, but it is my impression that he has been reassigned. Ask the Army locator. There will be a phone number in the Pentagon phone book for the Army officer locator. That's what you ask for. At least the Navy has one. You ask for the officer locator and they will tell you where Lieutenant General Hillman Dickinson is to be found. I'll try to remember. I think he's still in Washington. It seems to me I went to a cocktail party at his house at Christmas, so he couldn't have been reassigned in December, but he was going to be reassigned. I haven't seen him recently.
DeVorkin:All right. Let's move through things in ARPA that concern you directly.
Frosch:Vela did concern me directly, but it was not central.
DeVorkin:I mean, that you helped evolve and had a direct part in. We've talked about the organizational structure of ARPA down to your level.
Frosch:Now, I had a small office.
DeVorkin:Yes, I would like to know how the office was structured.
Frosch:I don't remember. It was small enough so that I didn't worry about it. I brought Harry Sonnemann on. I don't even remember who all the people were at this point.
DeVorkin:I'd like to ask you what your contact was with those who were in seismology, like Frank Press and Herb York during this period of time, and certainly your contacts with NASA.
Frosch:Okay. I don't remember any contacts with NASA. I remember going over and getting a briefing on something or other, but I don't remember on what. At some point in this period — it may have been then, but more likely it was later — I sat with a committee at NASA to look at the uses of Landsat, EROS, but I don't remember when. NASA could figure it out. It was for Len Jaffe.
DeVorkin:Why were you looking at it? What expertise did you bring to that?
Frosch:I think I was supposed to be the seismologist, but I don't remember. I think I was viewed at that point as a guy who had been in a lot of geophysical-type detection things, and was familiar with looking. Vela was looking out, and seismic instrumentation was looking in at the earth. It probably came in because of that connection. I guess it was the big measurement systems thing, since I built Artemis.
Frosch:Then at ARPA I built LASA, the large area seismic array. I worried a lot about sensing fields and phenomenon, and here was another way of looking down, so it was in that connection.
DeVorkin:Well, in talking about LASA directly, what was your contact with Frank Press, since he was in seismology at that time?
Frosch:That's when Frank and I got into closer contact. Frank and I had known each other slightly when he was at Lamont and I was at Hudson, maybe even when we were at graduate school, I don't remember. But we knew each other. And then in the LASA thing, Frank was one of the outside advisors on the whole seismology program. Now the seismology thing is interesting. Seismology was a central problem: could we develop enough knowledge and technology so you could tell an earthquake from an underground blast? And what were the detection limits of the underground blast? What I found was a bunch of very good people using essentially 19th century technology. It was better than that; they had galvanometers, and some electronics. But it was modernized smoked paper tape. Most of the seismometers that were around were using smoked paper. No kidding, clockwork and smoked paper. Now, it made certain kinds of sense. It was cheap and so on. There were still seismometers called wiecherts around, a weichert being a two-to-five-ton inverted pendulum with a pointer on the top that scratches its pattern in a smoked slide. Bob Sproul didn't believe me when I told him what I had been seeing so I took him into a seismic vault. It happened to be in Ottawa, because we were talking to the Canadians, and he came out and said, "I haven't seen anything like that since I went up in the attic of the physics building." The point about Frank Press was that he was one of the few seismologists who had the perception that it was going to be possible to make a revolution in seismology by changing the technology.
Frosch:And here was a motivation for changing the technology, which was more than the "science on a shoestring" that seismology had always been. Here was a driving, national reason to really update the theory and learn a lot, and build better equipment. I became the apostle of modernizing the equipment. I began to suggest, from my previous experience, to build an array. If you want to really have detailed detection and direction and so on, you build an antenna. Well, it was the same thing. I got the argument that it's incoherent and you don't appreciate how complicated it is, and the noise is coherent, and a lot of stuff that I didn't quite believe. One group of people, principally led by Hal Thirlaway in Britain — he was the head of the UK group — had already concluded that the way to deal with this problem was not to sit over the explosion, but to be a big distance away. Because you then used the paths that went deep in the earth, and they were simpler and cleaner.
DeVorkin:The P and S waves.
Frosch:Yes, they were going to be much simpler and cleaner. They had already begun to think about an array. So the world neatly divided into two groups of people. There were the people who thought that improved equipment and better arrays and so on would be helpful. Then there were what I call the classical seismologists, although that's unfair to them in a way, who felt that they knew as much as they needed to know. Their attitude was, "We're not building this kind of equipment, because it's no damned good." There were a few small commercial sidelines of people who were making their living out of having built the best available equipment, who didn't want that applecart kicked over. There were the Air Force people who felt they had a hammerlock on this and didn't want ARPA to get into it; and there were a lot of outraged seismologists who didn't like being told by somebody who wandered in from underwater sound, for God's sake, that they had missed the boat. Frank was the biggest supporter. He consciously felt that it was worth trying to do some of this, not only for the nuclear test detection, but because it changed seismology. Now, it's hard to say how the change occurred. We were pushing; at the same time, the computer business was improving, and other things might have changed it, anyway. At any rate, we caused a big ferment in the seismology business. Some of it was hinged on whether it was worthwhile, whether you could even make a big seismic antenna work. Then there was the question of whether it would be worthwhile doing it. Again, the same old set of questions: are the waves going to be coherent over an appropriate size? Is the noise going to be coherent or incoherent, and so and so forth.
DeVorkin:You received the Flemming Award in 1966. Was it for establishing a new detector system in seismological arrays?
Frosch:Yes, it was all of that set of contributors, but it probably rested in LASA more than anything else. The group that was doing this was Thirlaway, whom I mentioned, and his group, who were a theoretical part of the thing, myself, Harry Sonnemann, who had come on board and was really the construction engineer in a way, Paul Green and the group at Lincoln Labs, and Carl Romney. Romney was then at a thing called AFTAC, Air Force Technical Applications Center. He was the conservative seismologist in the lot, rather thought it wouldn't work but was convinced to cooperate and help. We approached it as an array antenna design problem. Just like Hudson Labs in a way, we got the guys in and sat around and figured out what it was we wanted to build and how to build it. We ended up with a design that had 200 seismometer vaults in it, that is, 200 instrument locations with a location in the ground. There were twenty clusters of twenty. There is a "Proceedings of the Royal Society" paper, and there's a Spectrum paper, and maybe a couple of others.
DeVorkin:I have for that period, "Instrumenting the Sea Floor" for IEEE Spectrum in 1964. And Green, Frosch, and Romney, "Principles of Experimental Large Aperture Seismic Array, LASA," Proceedings of the IEEE.
Frosch:Okay. There's also an article in the "Proceedings of the Royal Society."
DeVorkin:We don't have that.
Frosch:It is probably the same authorship, maybe including Hal Thirlaway, but I'm not sure.
Frosch:The IEEE one will, I think, give the design. Let me say parenthetically that it was successful enough so that there have been at least two, if not three others built. There is an Alaskan one, and there is a Norwegian one.
DeVorkin:What I'm getting at is that the Washington Post in '66 stated that you were awarded the Flemming prize because you had been responsible for the first new improvement in nuclear test detection. And I was just wondering if that was LASA?
Frosch:Yes, that was LASA.
DeVorkin:Okay, fine. Because it didn't specify what it was.
Frosch:Yes, but that is what it was. We produced this design, and obviously, this is a large object, twenty clusters of twenty.
Frosch:The outside diameter of the array is 200 kilometers in diameter, so that it more or less occupies the eastern portion of the State of Montana. The problem was standardization of seismometers. It's obvious that you can't run this with smoked paper. It's going to have to be operated as a wired-up antenna, and the only way you're going to do the signal processing and the management is with a computer.
DeVorkin:Yes. Aren't P and S waves of much higher frequency than in underwater?
Frosch:No, much lower.
DeVorkin:Are they lower than 30 cycles?
Frosch:We're talking down around 1 cycle, from a 10th to 10. That's why it's so damned big. You've got 200 instruments pumping out data. They are all over the place. How are you going to run 200 seismic vaults? How are you going to know whether the instruments are in calibration and the power is right, and all that? So we developed this concept of a completely wired-up array in which you never went into a seismometer vault after it was constructed, unless you knew there was a problem. This was largely Paul Green and the Lincoln group; the array was to be managed by the computer, and operated by the computer, and the data collected and processed by the computer. This was kind of a wrench to a lot of the seismologists. Frank Press was applauding and pushing us and so on, but the other guys were worried. You know, "you're not going to go in and change the paper? How are you going to know it's operating?" and so on. Everybody's nose was out of joint because here are these guys from Lincoln Labs in the radar business, whom I have employed to go be a major contractor and run the electronics design, and so on.
DeVorkin:LASA was obviously a very big project. ARPA was contracting with Lincoln Labs and other groups to perform this project. It must have occurred that within ARPA, you made the decision — fundamentally a policy decision — to go with new technology. Therefore, that determined the typesof people you were going to subcontract with, those you would not subcontract with, such as the classical seismologists groups. Is this really the scenario?
Frosch:I decided that I was going to push to do this. I sold it to Bob Sproul and Charlie Herzfeld and Harold Brown.
DeVorkin:So you had enough technical expertise within ARPA itself to make these decisions?
Frosch:Yes, I had enough technical expertise to make that decision, given that I had people I could trust, that I could lean on for details. Bob Sproul and Charlie Herzfeld and Harold Brown are very good physicists. They are not so much engineers; they've been around a lot of equipment, and, yes, it all made sense. Frank Press and some of the other seismologists said, "Yes, this is worth a trial. It's worth the money." They were all the usual guys who said, "Why, for seven million dollars, think of all the smoked paper I could buy." And there were lots of other complications. ARPA had just built the world-wide net of standard seismic instruments, which was a perfectly sensible thing to do. This was not really competitive with it, but it was perceived as that. There were all of those things. We felt perfectly confident that we knew what we were doing. Seven million dollars or eight million dollars for an instrument was, after all, not extraordinary on the ARPA scale. We were just in the process of launching two satellites. I don't remember what the Velas cost, but it must have been fifty million apiece by the time you got them launched. This was gigantic from a kind of "down in the basement" seismology, but not gigantic from the point of view of defense instrumentation. It depends on which world you were in, you see. Even from the underwater sound point of view, which was not a rich community like the nuclear or the space community, bidding on a seven million dollar project was a small fraction of a destroyer. You were kind of accommodated to this. If you were a university seismologist, running on fifty thousand a year for budget and graduate students in half the department, you would see things differently.
DeVorkin:That was a lot of money for them.
Frosch:That was a lot of money for them. You see, at that point, ARPA was supporting seismology in the U.S. That is the only way to describe it. ARPA was the principal support of every single seismologist in the U.S., bar two. I think there were two people at Fordham who were not contractors, for some reason of their own, which I never found out. So it was the whole seismology community that you were talking with.
DeVorkin:No matter what their interest — continental drift, this and that and the other thing? They were applying to ARPA for funding as opposed to NSF at that time?
Frosch:Well, my recollection is that every private seismic interest in the U.S. was supported by ARPA.
DeVorkin:You say, private?
Frosch:Private, yes. There may have been government ones. The Air Force had a net and so on.
DeVorkin:I see what you mean, yes, but university?
Frosch:The Geodetic Survey had a net. All the university seismologists were.
So we set out to build this thing, and we had to get seismometers designed and built. That was really an adaptation from existing instruments. Nobody ever built this scale of seismic instrumentation, cabling and wiring. Then we discovered we had a lightning problem, because lightning would induce surges. I was used to underwater cable. You had all that nice conductive salt water — you don't get lightning hitting the bottom of the ocean, even though you may get a signal from a lightning stroke. The Eastern Montana plains is where we decided to put it, after looking around and discussing whether it was all right to put it on sediment, or on solid rock. And Eastern Montana has heavy thunder storm activity. The problem was that with 100 kilometers of cable, you're getting up to a size where it becomes a good antenna for receiving the electromagnetic waves from a lightning bolt anywhere in the area. If a wave starts down one of these cables, when it gets to the termination, you can have enough voltage and current to blow up the seismometer. The question was what to do about this. We thought that the telephone company had obviously solved this problem, until we discovered that they had only a fix for the problem.
It wasn't one we could use, because they had a different power termination system, and so it was not a real solution. Harry Sonnemann, bless him, solved that one by figuring out that the key was to make sure the line was balanced. if it stayed balanced, you could sense that this surge was coming. Thank god for transistors. We put in a balanced transistor circuit that decided when the voltage was too high, it would cut the line before it got to the seismometer. It didn't do this for other noise fluctuations. Anyway, he got that problem solved and demonstrated. Then we had the problem of building the blasted thing after we had it designed. Obviously, the way to do it is to use the local telephone system as much as possible. It turned out we didn't. We used them to do some of our work, because we needed telephone lines to carry the data from certain central points into Billings, Montana, which is where we put the central station. After that, we decided we had to bury the cables. We had to negotiate for easement rights for all the cable and for permission to put vaults in on people's property, etc. I don't remember who did all of this. We needed electric power, and it turns out that's an area of rural electric cooperatives and rural telephone cooperatives. We were dealing with 15 rural electrical cooperatives in order to get power into the various vaults.
DeVorkin:How secret was this?
Frosch:This wasn't secret at all; thank god, no classification. All perfectly open. And I think there were eleven telephone cooperatives, and I have no idea how many private property owners. We had to find out the most efficient way to lay thousands of miles of cables, using cable trenching equipment and so on, and getting easements. There was a contractor who did it. There were all sorts of funny problems. We were losing time in construction because it had rained and the crews with the trenching machines were having trouble with rattlesnakes.
Frosch:There was the guy who ordered us off his land, even though we had a signed contract for the easement. He said, "I was having breakfast yesterday and I looked out of my windows, and I could see people on my land. And I don't like to see people on my land while I'm having breakfast." We untangled that one. Anyway, this thing got built, partly with luck and partly with good management. It was built on time and on cost, which was phenomenal in the Defense Department at that time, or any time. We were able to do it this way because we operated like a small business, in essence. Although we were embedded in the Defense Department, we were ARPA, and we had our own procurement people. If I had a procurement problem or a legal problem, I walked down the hall to talk to Herb Test. He was the guy who ran that for us, and was experienced and knew about it. He was not, you see, a clerk at the end of a long line of clerks in a system which had been rigidified, so that it could deal with 50,000 pairs of shoes. That tends to be the problem, you see. He wasn't dealing with 50,000 pairs of shoes. He was doing procurement for ARPA. ARPA's pride was that if somebody had an idea today, and we all thought it was a great idea, he could have a contract tomorrow, I mean, literally. We actually did occasionally do this. Somebody had an idea. We said, great, and he had money authority in the morning. So we had a way to do that.
DeVorkin:This is a very important thing to understand, I think, because it's so different now.
Frosch:Right. This was because it was a small operation; it was special. It was not in the Army or the Navy or the Air Force. My lines of communication were short. If Paul Green had a problem he could call me. If any of the contractors had a problem, they'd call me or they'd call Harry. Harry could walk in. We could talk about it. There were others. My memory is not so good on the people and the names.
DeVorkin:That's all right.
Frosch:Anything that would be a refresher would help. So if I had a problem, I could talk to my own procurement guy, I mean, literally. I'd walk down the hall and sit down and say, Herb, I've got a problem. We'd go solve the problem. If I had a problem we couldn't solve, I could walk across the hall. I was in the D-ring, so I'd go around to the E-ring in the Pentagon, and talk to Bob Sproul or Charlie. If necessary, he could say, "let's talk to Harold." and pick up the phone: "Harold, we've got a problem. Can we come see you?" Harold would worry about getting to McNamara. We never really had to do any of that, but you know. that was the general flavor of it, and I knew I had money. I knew where my money was. I could handle it myself. Somebody might say, "You can't do that, Paragraph 23 of the Procurement Regulations says you can't do that." We would find a way to do it, tell them to look for another paragraph. By then I knew enough about ASPR, Armed Services Procurement Regulations, to know that, in fact, there is an escape clause. There's a whole section on R and D which says, for R and D purposes, if necessary, nothing else applies but this. Most of the procurement offices didn't know that. We had that clutched in our little hands. I knew how to operate in the system, or knew people who had done this before and knew how to operate in it. There was also another way. If you couldn't do it through the government system, then you could quickly tack a task onto one of the contractors who you trusted and understood, and make him do that, too. His practices might be more flexible in this regard than the Government's. Obviously, you had to trust everybody's honesty and competence and so on. It was a small enough group so you could do that. The lines were short.
DeVorkin:That's a very important point.
Frosch:You knew who all the people were. If there was a crisis. everybody would fly in or you'd go some place and you'd sit around the table and deal with the crisis. You had enough authority in your own hands. This is relevant to NASA. It's very important that NASA is an independent and not terribly large agency. It's got 22,000 people, but it's organized around the centers. There is an organizational point here. If you need to — and we have done it — you can get all the center directors and all the directors, of the major offices in one room at one time. They all know each other. You can get them on the phone at one time, and clear things up in a hurry.
DeVorkin:Did you have any crisis in particular while at ARPA that required people getting together quickly?
Frosch:Yes, but nothing that was such a big thing that I remember. We were doing it all the time. You know, there are always crises. I mean, the cable supply supplier is on strike, etc.
DeVorkin:Yes, that sort of thing.
Frosch:I don't remember any major ones.
DeVorkin:Did your success with LASA bring you to the position of Deputy Director of ARPA?
Frosch:I suppose. I don't know.
DeVorkin:Did you apply for that position?
Frosch:No. Bob Sproul was going to leave. Everybody knew that Bob Sproul was going to leave. I don't know how it was arranged that Charlie was going to take over. Charlie called me up one day and asked me if I wanted to come over and talk. And I went over and talked and he asked me if I wanted to be his deputy. That's about it.
DeVorkin:That's a pretty short line.
Frosch:Yes. Now, what machinations the system went to so that I was the legal candidate, whether there was a search or what they had decided to do, I don't know. But it was pretty straightforward. There were other things. On-site inspection was a perpetual question. Somebody decided that this has to be attacked as a social problem, etc. I guess that it was Art Barber later, who was the Assistant Secretary of Defense for ISA, International Security Affairs, and there was a State Department thing about on-site inspection. There was a thing called Project Cloud Gap, which was going to do experiments in on-site inspection. After a little experience with it, I promptly renamed it Project Loud Crap.
Frosch:Loud Crap, because they were doing some of the dumbest things. There was an argument about whether you could determine whether troops could move in a surreptitious way without alerting on-site inspectors. They ran some dumb exercises in which they put people on street corners in Albuquerque with binoculars to see whether they could tell whether a bunch of tanks came through town, some dumb thing like that.
DeVorkin:Were there reconnaissance aspects of Vela, direct visual reconnaissance?
Frosch:No. This was happening in another arena, and it's sort of an introduction to what I am about to say. This was not in ARPA. This was happening in the State Department.
DeVorkin:So ARPA never did anything with the visual reconnaissance in the Landsat type?
Frosch:No. none of that. I knew about Landsat and its genesis, because we were keeping an eye on all possible systems for combining data. After awhile, I got cleared into those in the Pentagon who knew about intelligence systems. This must have been shortly before I became Deputy Director. I got involved in that community of the intelligence watchers in the Defense Department.
DeVorkin:Is there anything you can tell us about the problems and development of Landsat because of its similarity to military?
Frosch:No, I really wasn't involved in that. Anyway, I got a call from Bill Godel one morning. I had a habit of getting in early, and Bill calls me up and I'm in. When I say early, I mean like 6 a.m., or 6:30. I guess I had a paper to write or something. I couldn't find any time in the office, so I got in very early. I have always been much better in the morning — I fade out in the evening. It's a little worse as I'm getting older, but I do pretty well in the morning. He called me early and said, "Look, there's this question of on-site inspection, and there may be some money that we can get our hands on for a program, if we can act fast. Do you think there's anything worth doing?" I said, "Yes. I think there are some things worth doing." (although I must say I don't now remember what they were). He said, "Can you give me by noon a program proposal with a budget for doing something sensible in on-site inspection?"
Frosch:By noon. I said, "yes," and I went back to the office and I sorted out my thoughts. Out of my head, I constructed a several million-dollar program. It was perfectly legitimate. I did have some ideas. I thought about it, what's worth looking at, what are the concealments. I had become sort of an imaginative concealment expert. There were so many people around saying on-site inspection is easy — you can't conceal anything. I thought it would be useful to have somebody who made the assumption that it was possible to conceal things, and amused myself by inventing elaborate schemes for fooling people. With that as background, I sat down and I devised it. I do not now remember what was in the program, or how big the budget was. It was probably a couple of million dollars a year. I wrote it out and got it typed by noon; it was very short, you know. It was something. By the end of the day, Charlie Herzfeld called me, or somebody called me up, and said, "Okay, you're going to have to do it." Somebody had taken it to Harold, and he says, "Great idea." He took it to McNamara and he said, "Fine; go do it."
DeVorkin:Within the day?
Frosch:Within the day. So we tooled up. Whatever it was we did, it made such a big impression on me that I don't remember what it was. But there was some operation like that. That's how fast it could be done. A funny anecdote. There was a period in which I found myself doing what I can only describe as sending myself memos via Bob McNamara's in-basket. McNamara developed some questions or other about nuclear test detection, something happened and triggered a question, and he sent Harold a memo. And Harold said, "We1l, there is this guy in ARPA who knows about that," so Harold sent me a memo, with a copy of it to Bob Sproul, or through Bob Sproul. I get this memo: what about thus and so and thus and so? So I write a memo to the Director of Defense Research and Engineering with a technical page or two explaining what it was about. This is the reconstruction of what happened. Harold took it, endorsed it and sent it to Bob McNamara. And Bob McNamara said, "Oh. yes; that really makes an international security problem." He sent it to Art Barber or whoever it was, and Art said, "I wonder about such and such. Who can answer that? Oh, I guess I'll send it to Harold," and so Harold sent it to me, and says, "Hey, there's this other question." After, that it went back to Harold and to Art Barber. When this happened twice, I realized I was both asking and answering questions and I went to Harold Brown told him that this has got to stop. The circle was Harold Brown to McNamara to Barber and to Brown, to Frosch, to Brown, to Barber, to McNamara to Brown.
DeVorkin:No wonder I got the names mixed up. That's amazing.
Frosch:Yes, it was; it was pretty funny.
DeVorkin:How did you like working at that level?
Frosch:Oh, it was fun. I enjoyed it. I briefed McNamara once or twice. I briefed the Joint Chiefs once pretty early on which was kind of fun. I think they decided, you know, we'll throw him into that cage and see what happens. I was asked to go down and brief the Chiefs on nuclear test detection.
DeVorkin:This was after you were Associate Director?
Frosch:No, this was while I was doing nuclear test detection. The spirit of ARPA, as will be seen later, was: you got a guy, you trust him; let him do the briefing. If the Joint Chiefs want a briefing on nuclear test detection, send a nuclear test detection expert. Charlie Herzfeld went down to watch this, I think. Bob must have been away. That was fun. I went down there, and I made my pitch. It was very impressive, all the Joint Chiefs and all the staff. Who was the Chief of Staff of the Air Force, big, cigar-chewing — Curtis LeMay.
Frosch:Curtis LeMay was pretty hostile to all of this. He took all this nuclear test detection as a kind of a threat to the strategic thing.
Frosch:It was a great cause celebre, because he asked me some terribly hostile questions. He said something like, "I deduce from this, thus and so, that nothing will work." And I just said, calmly, "Well, you know, that's a very reasonable question. General, but it isn't quite like that. Let me explain what the thing is." So I launched into this calm explanation. And he quietly subsided, because it clearly wasn't like that. It was a great cause celebre, because apparently, I was supposed to collapse at that point. When LeMay asked you a hostile question. chewing on a cigar, you were supposed to collapse. You weren't supposed to say, "Well, not quite that way, General." You were supposed to collapse or treat him like an idiot, both of which were fatal.
DeVorkin:Ah, yes, yes.
Frosch:It was a great celebre: our boy did well in the barrel today.
DeVorkin:That sounds great.
Frosch:That was interesting. I enjoyed that. It was a lot of fun. You had a feeling that you were doing something, that you could get something done.
DeVorkin:Let me ask you some general questions about ARPA and possible projects related to ARPA. ARPA did engage contractors from outside, as you said, Lincoln Labs, and places like that?
Frosch:Yes, lots of places.
DeVorkin:Did you engage individual consultants from universities and industry who worked with you?
Frosch:Oh yes, oh yes. Lots and lots.
DeVorkin:And you were able to fund them as quickly as you would be able to fund internal projects?
Frosch:Oh yes. We had standing arrangements for doing that. It was automatically written into any such task that anybody who was doing research for us was also a consultant to be called on when required.
Frosch:So, this whole university community was a running pool of consultants.
DeVorkin:And they needed security clearances as a function of what they were doing?
Frosch:Yes: if they needed it, they had it.
Frosch:If they didn't need it, they didn't. We could deal with that pretty rapidly. We had our own guy who was responsible for security, even though all the clearances, obviously, went through the standard military department thing.
DeVorkin:Yes, right. Did you have any contact with JASON, the JASON program?
Frosch:Yes. In the first place, I knew a number of the JASON people, anyway, from previous contacts. JASON belonged to IDA, the Institute of Defense Analysis at that point. It now is operated by Stanford Research Institute. But then it was IDA, and we had good contact with IDA. I'm glad you reminded me. IDA did work for ARPA, on LASA and related things. I went and talked to JASON about what we were doing, because they were interested in knowing whether there was anything useful they could do. I don't know if you know about JASON.
Frosch:JASON put on consulting contract a very, very good bunch of physicists so that you could throw problems at them. Originally it was physicists. It's now broader. It's got engineers and other kinds of scientists and technologists. They select their own problems.
DeVorkin:You throw them at them and they pick up what they want to?
Frosch:Yes, somebody would come in and say, "Here's my problem. Here's what I'm doing, and here are the problems." It might be a mathematical problem, or a theoretical physics problem. It was all theory, nothing but theory. There was no experimental work generally, although they would occasionally suggest an experiment.
Frosch:They might take them up, but it is their choice. Somebody will say, "Hey, that's interesting. I'll think about that." and go away and write a paper on it.
DeVorkin:Do they actually physically meet?
Frosch:They do physically meet. They meet several times a year, and they'll have a set of briefings and discussions. Sometimes they have a summer study, in which some or all of them will get together for several weeks and work on some problem. I know a bunch of them. Some of them were professors I had had. Charlie Townes was there. Henry Foley is a member of JASON. Dick Garwin, whom I knew from Watson Labs. A lot of people who had been around PSAC were in JASON, and so I guess I talked to them about some problems. Nick Christofilos was in that.
DeVorkin:Oh. Christofilos, yes. Anything with the Vela program, the satellite part of it? Was there any need for JASON or for any consultancy?
Frosch:My recollection was that they were involved at some point.
DeVorkin:Earlier than when you were there?
Frosch:Earlier than when I was there, maybe slightly when I was there, but I don't remember really.
DeVorkin:Was Giacconi in on this?
Frosch:I don't think he's ever been in JASON. I don't know.
DeVorkin:But he was a consultant for ARPA?
Frosch:I think so. You see, when I met Ricardo again in NASA, I did not have any recollection of having talked to him. He's about my age. He could well have been involved, but I don't remember.
DeVorkin:Yes. We've got about five minutes, ten minutes left.
DeVorkin:That's interesting. I appreciate your comments on JASON, and the fact that it was IDA, and now it's Stanford Research. You already mentioned that ARPA was not involved in satellite reconnnaissance. Did you have NRL contacts and APL contacts as Deputy Director? You knew what ARPA was up to. What other things was ARPA doing at that time?
Frosch:Well, there was a thing on what I'll call counter insurgency social systems.
Frosch:Yes. It was a very sensible thing, and Thailand was the experiment. It's really a problem in development, in the economic and social sense. It's clear that parts of Thailand are ripe for all sorts of problems. They barely know there's a central government, except via taxation and other things. It is an undeveloped infra-structure. They don't have a police system that works, even for ordinary crime. They don't have any communications. So what is it that one could do that would be benign, in the sense of being useful as economic and social development? What would improve both their capabilities and their interest in being resistant to outside influence? It's not a matter of being our influence or somebody else's, but to make them into a more integrated culture with an identity. Some work went on in that. We tried all the funny exotic things. We had things like how do you make a vehicle that will wander around in the Mekong Delta over the paddy bunds (paddy bunds are the ridges separating the rice paddies) in the muck. You know, what is the Army going to use to move around in there? We had all sorts of exotic vehicles that people were building. There was an excellent program in computers that Licklider started. He is an acoustician; I knew him slightly. We had Ivan Sutherland, who now is on the NASA Advisory Council, because I got him on it. He's connected now with Cal Tech and JPL, although he may recently have gone off independently. He was a brilliant young guy in the computer business, the first guy who actually built a thing called sketch pad. It had hardware and software, so that as you drew, an instruction was sent to the computer. He did the first programs in which you could sketch a bridge truss and indicate the stresses and the sizes of the members, and the computer would absorb that and redraw it for you, and compute your problems.
DeVorkin:Oh. that's very interesting.
Frosch:He was doing all of that. ARPA was extremely interested in that kind of computer development. There was a lot of early work on computer data bases and input, output devices.
DeVorkin:Who made the decision at ARPA to get into that?
Frosch:The director did. I think that had been done before. That was probably Bob Sproul or Jack Ruina who had decided that. I'm trying to remember the other programs. There was, of course, ballistic missile defense. ARPA was the BMD outfit.
Frosch:We had a lot of contracts for SPRINT, a very fast rocket, and for the various fancy radars. All that stuff was done before it was turned over to the Army.
DeVorkin:What about Project Mac? What was that?
Frosch:Oh. MAC. That was the computer.
DeVorkin:Was there a certain type of computer analysis that was being generated through ARPA support at that time? It wasn't time-sharing, was it?
Frosch:Yes. The early time-sharing was done by ARPA, and then ARPANET after my time. The central theme was the development of time-sharing, and the idea that the computer didn't have to be in any partictilar place in relation to the people who were using it. If you had a big machine, a lot of people could use it together, and so on. That was all being generated.
DeVorkin:The decision that ARPA was to do this was pretty much made before you had gotten there?
Frosch:Yes, as I recall. That was an ARPA decision, an ARPA proposal, and they set off to do it. You see, ARPA was supposed to do things that were going to be important to the Defense Department, things that didn't fit in any service at all, or was multiservice at the inception. Most of the projects were risky enough, or far-out enough, so that it was clearly not going to be easy for a service to take it up until there was something established.
Yes. So clearly, you were creating new interdisciplinary activities. Sam Koslov, who was a graduate school classmate, later was a contractor to ARPA. I got him onto the ARPA staff. He and I examined the intelligence business, not reconnaissance, to see if there were any worth taking up — bug detection, etc. I invented a thing that I called the generalized Sherlock Holmes problem: given an object, what can you find out about it, its origins, etc. I'm just running out of tape.
IEEE Spectrum, letter to the editor, ca. 1975. (Frosch letter dated 5/6/75).
IEEE Spectrum, 6, 24–25, 1969.
A. Berman, C. S. Clay and H. B. Sherry. Report AD 3129672. Secret. April 25, 1956, 31 p.
1. Frosch, R. A., Green, P. E. Jr., (1966) “The Concept of a Large Aperature Seismic Array.” Proc. Royal Soc. A., 290, pp. 368-384. 2. Green, P. E., Frosch, R. A., and Romney, C. F. (1965). “Principles of an Experimental Large Aperature Seismic Array (LASA)., Proc. of IEEE, 55, No. 12, p. 1821–1833.
1. Frosch, R. A. (1964). “Instrumenting the Sea Floor.” IEEE Spectrum, 1. No. 11, P. 101–104; 2. See above, reference 4 #2.