Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
Please contact [email protected] with any feedback.
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.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Norman Ramsey by Paul Forman on 1983 July 12,
Niels Bohr Library & Archives, American Institute of Physics,
College Park, MD USA,
For multiple citations, "AIP" is the preferred abbreviation for the location.
Developments of the technique of separated oscillating fields and the atomic clock. Move to Harvard University from Columbia University and Brookhaven National Laboratory; work at Harvard concentrating on the first molecular beam magnetic resonance apparatus, doctoral thesis of Harwood Kolsky; Jerrold Zacharias and the cesium beam clock; Brookhaven Molecular Beam Conferences (beginning 1947), significant developments in resonance. Also prominently mentioned are: P. I. Dee, Harwood Kolsky, Polykarp Kusch, William Aaron Nierenberg, Pendulchron, Ken Smith, John Hasbrouck Van Vleck, Robert F. Vessot, Earl Wilkie; Brookhaven National Laboratory Molecular Beam Conferences, Fort Monmouth, Frequency Control Symposium, National Science Foundation (U.S.), United States Army Signal Corps, United States National Bureau of Standards, United States Office of Naval Research, and University of California at Berkeley.
This is an interview with Prof. Norman Ramsey in his office at Harvard on the afternoon of July 12, 1983. My thought was that I couldn't do anything as ambitious as an oral history; my focus is very narrow.
That has an advantage in going into more detail.
I would like to cover two developments especially: first, the technique of separated oscillating fields; second, the question of a better atomic clock, with the ideas of broken atomic beams, and coming then to the hydrogen maser. Could we begin with your coming to Harvard and having to wait a little while before you became active experimentally, and somewhere in there this idea, as well as other theoretical ideas...
Basically I think that is a good place to start on it. I had been both professor at Columbia and head of the Physics Department at Brookhaven Lab. I was then offered a position at Harvard, which I initially turned down. I later changed my mind and decided I would come, chiefly because of a rough ride on the Long Island Railroad. I initially thought that I had a very good set-up, being both at Columbia and at Brookhaven, although the physical separation of the two institutions involved commuting time, I thought that I would effectively use my time on the train for study. I was at that time trying to understand Wigner's book on group theory, which was then available only in German. But with the combination of the German language, the subject of group theory, and the Long Island Railroad — on which I was trying to do the studying — it seemed to me I wasn't getting very far. So I changed my mind, called Van Vleck and asked if the offer at Harvard was still open. I then came to Harvard in the summer of 1947. I decided when I came that the key thing I wanted to do would be set up an even better molecular beam magnetic resonance apparatus. One of the ways you can improve it is, of course, to make the length of the uniform magnetic field and oscillatory field region much longer. Essentially it is an application of the Heisenberg uncertainty principle, E. t~h. If one makes t bigger, E gets smaller; one obtains a narrower resonance. This was the basic idea. But I also knew there was going to be a problem in it. Indeed it really had me worried while I was making the design and setting a graduate student to begin constructing an apparatus with a meter- or meter and a half-long field, when the longest we had previously had was about 15 centimeters. The problem I knew was the following one: even when we had gone to 15cm we didn't gain as much in narrowness as we had expected. The reason was that the magnetic field wasn't uniform. Indeed no one knew how to make uniform magnetic fields of that length. Therefore the molecules, in passing through the 'e' region, were subject to different fields. Thus although the intrinsic width of the resonance was narrow, since the resonance was at one frequency in one part of the magnet and a different frequency in another part, what one got was a superposition of all these, which could be an even broader resonance than before. Later, when we actually got the apparatus going, that proved to be the case: a single oscillatory field was much worse than it would have been with a shorter apparatus. Knowing that this problem was going to come up, I was worried. Most of my thoughts were directed to inventing a way to make the magnetic field more uniform. I had some good thoughts in that direction, involving a magnetic filter and a very high permeability pole tip, which worked out reasonably well. But even so, it wasn't going to be good enough. So there was really this underlying worry nagging me: here was this apparatus that seemed to be going along pretty well, but which also was going to be something of a failure as an attempt to get narrower lines. At this same time I was giving a course on physical optics, the last time this course was given at Harvard for some years. I was using Max Born's textbook, Optik, also only in German at that time. And with this worry [about the uniformity of the magnetic field and the width of the lines] in the background of my thoughts I came to the point of having to give a lecture on the Michelson stellar interferometer. [March, 1948] When I was an undergraduate at Cambridge University in England, one of my professors — I think it was probably P.I. Dee — had a really dramatic way of describing why the Michelson stellar interferometer works. Suppose, he said, that you were looking with a 200-inch telescope at two very closely spaced stars, and it was really worth a lot to you to separate the two stars. If the light intensity was plenty, so that the problem was just a lack of resolution with that size telescope, then if you took a can of black paint and went to the telescope and painted out all the middle of the telescope, just leaving two slits at the edge, you would double the resolution of the apparatus. That is essentially what the Michelson stellar interferometer does, because the interference pattern between two slits is twice as narrow as that of an open aperture. Well, I thought this was a nice pictorial way to describe it to my class; I was giving this explanation to my class. In the middle of describing it, two things suddenly dawned on me. One is, obviously, if you paint the middle portion of the glass black, it doesn't make any difference what the quality of the glass is there. And so of course to make a Michelson stellar interferometer you don't actually take a 200-inch telescope and paint it black. You just make two slits at a distance. And, [secondly,] I sort of had a feeling: isn't there some way I could do the same kind of thing with our magnetic resonance experiment? Well, this was the idea but I didn't know quite what "the same thing" was. It took a little while to see how it would come to pass. It isn't after all, the stellar interferometer. It's actually really separation in time not separation in distance; and it's quite a different problem. It is really a three-dimensional problem of vectors, since basically an angular momentum can point anywhere in three-dimensional space; thus it is a more complicated problem than the interferometer, where you have a vector with just two components, and in fact if it's used with polarized light it's a just single component. Well, I did start working and soon got myself convinced that, "yes, there was something I could do in that direction." I then launched into quite an extensive array of calculations, which you discovered in my notebook, trying to see if it really was a feasible idea, or wasn't a feasible idea. I calculated what the resonance should look like. And indeed, it turned out to be that way. When we were building the apparatus, I mentioned this to, I guess, Kusch on one of my visits to Columbia; and he said, "we can get an apparatus going here and give it a quick try, just to see if it works" — which it did. He just did one quick look with a rather asymmetric resonance to show, "yes, the principle was quite good." I was pretty convinced it was going to be. I didn't see any way it could fail. We built the apparatus, that also had the following somewhat amusing characteristics: Obviously with this new method, and an apparatus that was roughly ten times the length of the previous apparatus, and with the ability to take full advantage of that length our lines were at least ten time narrower than before. We also had better means of taking data. We also realized we could do interesting things at lower fields as well as at high. Thus in a number of ways we could improve. It was clear we could probably do, relatively easily, a factor of 100 better than the previous work with hydrogen and deuterium. These were particularly interesting molecules to study, because they were the source of the discovery of the deuteron quadrupole moment. Now we could measure the interaction very much better. But building an apparatus from the ground up, with a couple of new graduate students, in a totally new environment, always takes longer than one expects. My students had predicted that they would have the apparatus running, with the beam, by December '48. I can't recall the dates for sure; it was after I had been here a year or so; probably December of '48. Well, come December of '48 and no beam.
It was on a spring day, around April or so, that I went down to the laboratory and found lovely crepe paper, red and green crepe paper, decorating the apparatus. My students were celebrating. They had now achieved the "Christmas" which was the predicted date of operating. And it was operating, but it was also quite clear that there was still some way to go before physics results were obtained. This was, of course, right after the war. Many of the students had been in the war a long time, particularly Harwood Kolsky, who was my first graduate student on the molecular beams here. He was married, had two or three children at the time. A very nice guy, he obviously had worked very hard and was doing a great job of building the apparatus. Everybody wanted him to get his degree as fast as he could. He even had a new job waiting for him. But [to get a thesis] he had to get the apparatus running. Time went on, with feverish efforts. Finally we got it going, but we still hadn't gotten any resonance and done anything. By the Harvard statutes one is supposed to have his thesis submitted to the Ph.D. examining committee at least a week before the exam. The exam has to be at least a day or two before the report was due. Well, he had good friends on the committee. The committee agreed they would have the exam the day the reports had to be in, and the department agreed to it. A week before the scheduled exam there was still no resonance. By this time Kolsky had written all of his thesis — and I had even approved it — except for filling in the little section on results. Still we didn't cancel the exam. Finally the day before the scheduled exam, which was on the last possible day, the apparatus finally started giving resonances. And so instantaneously things were going to be great. He ran a couple of resonances, and then his associates ran it while he was filling in the data on his thesis. The thesis had already been submitted to the committee, missing the section on results. They had read that. The next morning, I think, having stayed up all night, and not slept at all, he went into the exam with beautiful results and passed in fine shape. To the best of my knowledge Kolsky holds overwhelmingly the record at Harvard for the shortest interval of time between getting the first bit of data in his thesis and actually getting the degree awarded. I think it was a total of maybe three or four days until commencement. It went into the printed program the same day he had the exam. I certainly have never heard of anyone who had even come close to that, even within a factor of two.
That was, the basic fundamental invention of the separated oscillatory field method. Then there came a period, as always, when we were trying to get it going better. And we were worrying about ways in which the resonance could be distorted. That particularly came up as a sort of independent development a little bit later. We were using the method in the lab here, and learning various things we had to watch for to make sure we had no phase shift between the two separated oscillatory fields, because that can give an apparent frequency shift. The next major development was actually in conjunction with phase shifts, in collaboration with one of my next thesis students, Henry Silsbee. I had the thought that we could make some real gains by deliberately putting in phase shifts between the oscillatory fields. In particular, if instead of turning the oscillatory field on and off as we had done in our previous experiments we deliberately made a shift from 180 ° to minus 180 ° and subtracted the two curves, we effectively doubled the intensity, gained a factor two with the same amount of data taking. And that was a very real help. We worked that out. We also realized that you could, for some purposes, do better to use not 180 ° phase-shift, but a 90 ° phase-shift. That gives a dispersion-shaped curve, i.e. instead of having a peak of the curve at the resonance frequency, we had the steepest slope of the line at the resonance, which is often advantageous to determine its position. So that was, I guess the next major development in the technique. I returned to these questions at a significantly later period when the National Company was working on atomic clocks using cesium. Their work was started by Jerrold Zacharias, who also has an intriguing history with a failed experiment with some wonderful results. Zacharias had a very interesting idea for getting a much sharper resonance, an idea that rather depressed me when he told me of it because I wished I had invented it myself. His idea was to take very slow molecules, shoot them up vertically, let them fall under gravity. It is well known that anything from a molecule to a baseball takes about a second to go up and come down a few feet or so. Therefore, one should get a fantastically sharp resonance. Our molecules were ordinarily in the apparatus for only a tenth of a millisecond. In this way he would get a full second, a vast improvement. Sadly, this experiment didn't work. It turns out that in the nonequilibrium configuration where the molecules are emerging from the nozzle there was enough scattering so that the very slow molecules got scattered out, and the fast ones didn't get scattered down. So it failed as an experiment. But in the process of doing the experiment, several interesting things came out. He began to realize he could do an awful lot just using the ordinary cesium beam. If you engineer it well, you can make an atomic clock, even one of commercial usefulness. So he got together with Vessot and Arthur McCoubrey and Holloway, to develop a commercial cesium beam clock, which they did effectively. A typical improvement they made: where one used to have to change the oven every day, they put a year's or two year's supply of cesium in it without the necessity of opening the beam tube. But they also were having trouble. This resonance shifted somewhat. So it ended up by their asking me to help them out. Since I wanted to do some further work on the resonance theory, I agreed I would, if they would pay for calculation time on the UNIVAC I. That was then a super machine, we thought even though it was considerably less powerful probably than a current HP pocket calculator or a Radio Shack smallest model computer. Nevertheless, it was the best we had then and so I actually did a number of calculations of the effects of the various phase shifts and of changing the amplitudes. It was pretty expensive then. The calculations told us a lot about what we had to watch for, and I think did a lot toward making the atomic clocks accurate. The results of these calculations I eventually wrote up in a place that is unfortunately a little bit obscure, namely, in a special book that was written in honor of Otto Stern's 60th birthday. It also appears in the French JOURNAL DE PHYSIQUE.
Do you remember how you came in contact with the National Company? That was before McCoubrey was there?
At the time I did the calculations McCoubrey may have been there. I'm not really quite sure. Of course, I knew Zacharias quite well. We were collaborators together at Columbia. And through him I had met some of the people there. Holloway and Vessot had been former graduate students of Zacharias.
Did you know Daly?
Yes, I knew Daly. I'm not quite sure, but I think actually they came around. They were worried. They had a problem. They had already made one of these, and then the Signal Corps for whom they made it… Incidentally the Signal Corps made a very valuable contribution in that case by purchasing some fairly large number — 20 or so of them. That really is what enabled the commercial version get off the ground. Unfortunately, when we got the hydrogen maser first going, nobody was willing at that time to put out the money for a similar purchase, so that we had a much harder time getting started. But this was a great contribution made by the Signal Corps to the National Company. But then, there was a complaint that the frequencies weren't quite as stable as they were hoped to be. I think they came to me saying they had a problem; did I have any ideas. And so I was handing out free advice. Then suddenly realized that I had been trying to do a little calculation on this but couldn't afford the computer charge to do it. So they gave me couple of thousand dollars for computations.
The Signal Corps activity seems to have been led in good part by Fritz Reder.
Yes, that's right. Of course, the organization on the financial end that deserves the greatest credit for getting things started was ONR. ONR did an absolutely magnificent job. They had the field almost to themselves for many years after the war, until NSF got started. Even after that they gave a very high standard of support. Although their methods of support were research contracts, which could be a real nuisance if badly administered, they operated them admirably well. They did not interfere with things. It was ideal environment for research. They were very smart in choosing topics to work on, all in their interest. They really were the model on which NSF was based. And it truly kept NSF up to standard in the sense that as a nonmilitary organization they ought at least to do as well as ONR.
The Brookhaven molecular beams conferences were beginning at this time?
Yes, that's right. In fact, I organized the first one in the summer of '47. It was an interesting time, but it also had one rather disastrous effect: It was very clear at this time that there was a huge open problem — and I was thinking quite a bit of working on this at Brookhaven namely measuring spins and magnetic moments of radioactive isotopes. One could get radioactive isotopes in sufficient quantities to measure; it was a clear thing to do. I hired Bill Cohen and we did get molecular beams started at Brookhaven. But before launching into the big program on the radioactive isotopes, I thought it would be good to call a conference of all the people who had techniques for measuring spins and moments. In the period from 1945 to the summer of 1947 lots of things were happening in the resonance field. NMR had been invented somewhere in '46 by Purcell and Felix Bloch, etc. Microwave spectroscopy was just getting started. I don't think optical pumping was yet in, but there were several quite new ways of doing things. And then there was one older technique, namely molecular beams. Purcell was at this conference, and also Charlie Townes was there, and quite a few groups. All the people who had done molecular beams were there, except for one person I didn't know was doing molecular beams, Ken Smith, a graduate student working with Otto Frisch at Cambridge. They were, indeed, working up some apparatus, and had thought about doing radioactive isotopes. But they didn't know about the conference and I didn't know about them. So they were not invited, which turned out very fortunate for them. We had everybody describe what he could do, how well he could do measuring spins and magnetic moments. We molecular beam people had been measuring magnetic moments for some years, so we knew how hard it was. We knew all the problems that we were going to run into. We knew that each isotope was a research problem in itself. It came slowly. You didn't just gear up your apparatus and have the results pour out. But it could be done. So we presented our results: I presented some, Bill Cohen presented some. I think everything we did was absolutely right. What no one reckoned on was that the other people had not yet measured any radioactive isotopes apart from hydrogen, which had been measured by NMR. Hydrogen is the easiest thing to measure; it's got the biggest moment. As they had no difficulty, they were quite confident there would be no problem measuring the smaller magnetic moments. The microwave spectroscopy people hadn't yet measured any magnetic moments or spins, but they were sure they could do it. Well, the net effect was that the molecular beam people presented their views from a pretty realistic point of view. The others were all presented from, as it developed, a highly optimistic point of view. So the clear net conclusion of the conference was the way to do nuclear moment measurement was microwave spectroscopy or NMR. We with our molecular beam technique would just drop out and wait for the answers to pour out. Well, Ken Smith wasn't invited to the meeting, so he plugged merrily away. The first measurement of a radioactive isotope was in fact done by Ken Smith. Then it began to become clear that it really was quite difficult to measure a previously unknown magnetic moment by the other new techniques. After I was at Harvard less than a year, Berkeley tried to get me to come there to work on radioactive isotopes. It was quite a tough decision. Berkeley is really a very nice place, too. But I decided I preferred to stay here; and recommended that they get instead one of my first graduate students, Bill Nierenberg. Although Ken Smith did the first measurements, the one who really made a large number of measurements was Nierenberg. And it is still true that almost all of the new spins and magnetic moments have been measured by molecular beam techniques. The molecular beam technique has this huge advantage over, say, NMR, namely that one can broaden the resonance. If one doesn't know where the magnetic moment is at all, he can operate the molecular beam apparatus in a low magnetic field and have all the resonances close together. In a few hours one can search any region in which a moment can possibly be and get just as big a signal. With NMR, on the contrary, one has to have a very large magnetic field in order to get a significant difference in the populations of the states, i.e. in order to be able even to see the resonance. But then, as soon as one does that, he has to search over thousands of megahertz, and do so with a very narrow line, or else you get too much noise. That can take a long time. I was involved in a striking example of this. Rather early in the development of the shell model it became crucial to determine the spin of neon 21. There were two models which gave quite different values of the spin. Thus this was clearly an important question. The NMR and microwave spectroscopy people said they were going to measure it; so I said, "fine, we'll wait and listen." We waited about two years and still no results. So I got a sample of neon 21, and within a matter of a day we had actually measured both the spin and magnetic moment. Now, with the moment having once been measured, the NMR could measure it even better. They could get much better accuracy by getting even narrower lines. But finding it in the first place is really much more difficult with techniques other than molecular beams. Incidentally, one of my collaborators on our present experiments on neutron electric dipole moments and parity violating spin rotations at Grenoble is the same Ken Smith. He is now a professor, about to retire, at the University of Sussex. A month ago I visited him for about a week. I told his wife of the great contribution I made to his research by not inviting him to a conference. That made possible his first experiment.
In the conferences in the mid-50s undoubtedly Zacharias' molecular fountain device, as a way of getting an extremely precise clock, was appreciated, but for what purpose?
Even while I was still [a graduate student] at Columbia, we realized the possibility of a molecular or atomic clock and of a magnetometer. I talked a little with the Bureau of Standards. They thought it was a little complicated for a clock, and even more so for a magnetometer, as it was then [primarily conceived]. And we did discuss a relativistic experiment with an atomic beam clock at that time. I know that it was back then, because I remember saying, as we were discussing putting clocks on the top of a mountain and the bottom of a mine shaft, "I know what will happen; as the graduate student I'll inherit the bottom of the mine shaft while my colleagues, Zacharias and Rabi are up on the mountain top enjoying the view." As a matter of fact, I had that same experiment in mind for the hydrogen maser. I was telling Bob Pound about this while we were still constructing it. The next day he came up with the Mössbauer way of doing it. Of course, now it's gone back again to the hydrogen maser. This beautiful experiment of Vessot's is now the best determination; the shift is determined to a hundredth of a percent, which means the whole time measurement has to be good to parts in 1015 or so. We couldn't have come close to that [before the War], but I think we would have had a chance, perhaps on a mountain top, with a really good cesium apparatus, actually to have detected an effect.
John King recalls that at one of these Brookhaven conferences-probably in 1956, probably when Zacharias exhibited one of his first Atomichrons…
I remember him exhibiting one, yes.
…that you came in with an old-fashioned bracket pendulum clock which you had set down beside it, with a label, Pendulchron [?].
I had forgotten that. I can well imagine doing it, but I don't remember having done so. I'm glad to be reminded of this. Those Brookhaven conferences were very valuable. It was really Bill Cohen who carried them on. Although the first conference was primarily of molecular beams, we did have a few people representing other techniques. Thereafter they were really molecular beam conferences for half a dozen years. Then gradually they shifted to atomic physics conferences. In fact, the current international conferences on atomic physics, such as the one I went to last year at Göteborg, Sweden, is a direct descendant of these conferences. They started numbering in a later period; the numbers don't quite go back to this very first one, that to which we didn't invite Ken Smith.
Could we turn to the broken atomic beam? In your first publication in the Physical Review Letters, with Kleppner and Fjelstad, you mention that this idea was first presented at a Fort Monmouth frequency control meeting in early '56.
One of the valuable things that emerged from Zacharias's ventures was the recognition on my part of there being a frequency problem, that there was a use for even more accuracy, pretty much without limit. I was originally not aware of this need, and had not given much thought to it. So well before that time I was going fairly regularly to the so-called Frequency Control Symposium, I gave various talks proposing ideas. I did say something about what we were planning to do. How much I said I do not remember…? When I was looking for data books, unsuccessfully, I saw downstairs a whole array of the earlier proceedings of that conference. I probably gave it as an oral report. And they didn't at that time report discussions.
While it's still in my mind, although it's a step further back, all three of your first graduate students here speak quite gratefully of the help of the machinist, Earl Wilkie. I gather that Wilkie was around until shortly before you began to work on the Hydrogen maser.
He did most of our molecular beam instruments. He was excellent as a machinist. He did just fine work. In fact, we had good luck with all of our machinists, but he was the principal one assigned to us. It slips my mind why he left or where he went, but it was certainly a friendly departure, because we thought very highly of him.
Do you have any recollection where that pair of separated field coils which we have on exhibit fit into this development? It is clear they weren't intended for this high-precision apparatus that Kolsky, Phipps and Silsbee were building and worked on.
I was never sure quite what they were. We did build another molecular beam apparatus with separated oscillatory fields in an adjacent room. In fact, that apparatus ran until a year ago.
[Interview continued some minutes more after end of tape reached unnoticed]