History Home | Book Catalog | International Catalog of Sources | Visual Archives | Contact Us

Oral History Transcript — Dr. Gerson Goldhaber

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.

Access form   |   Project support   |   How to cite   |   Print this page


See the catalog record for this interview and search for other interviews in our collection


Interview with Dr. Gerson Goldhaber
By Ursula Pavlish
At Lawrence Berkely Laboratory
July 30, 2007

open tab View abstract

Gerson Goldhaber; July 30, 2007

ABSTRACT: Goldhaber describes the milestones and turning points in the SCP’s history: the discovery of the first supernova after three years, then discovering batches of supernovae at a time, the use of a 10 meter KECK telescope to get spectra taken. Goldhaber describes his tables of supernovae. He explains how two images are taken each time when searching for supernovae, to avoid hot pixels, cosmic rays, and asteroids in the data. Goldhaber’s discovery of a peak in the data, shown in Santa Barbara on December 14th, 1997. Comparison of this method to Goldhaber’s previous particle physics work. Explanation of tables he made of the data.

Transcript

Pavlish:

It is July 30th, 2007. My name is Ursula Pavlish. I am here to interview Professor Gerson Goldhaber in his office at Lawrence Berkeley National Laboratory. I would like to ask you, Professor Goldhaber, questions that fall into several categories. The first category consists of factual questions.

Goldhaber:

Yes.

Pavlish:

The second, on how science works, through the lens of this particular discovery of the accelerating universe. Now, you have participated in many discoveries in the past.

Goldhaber:

Yes.

Pavlish:

But I would like to ask you about the discovery of the acceleration of the universe, in context of your previous discoveries, to highlight aspects of how science works. For instance, on the relationship between theory and experiment, how experiments end, scientific visualizations, or scientific objects. Scientific objects are objects that are scientific or become scientific because they are studied scientifically. For example, the supernova would be a wonderful scientific object.

Goldhaber:

Yes.

Pavlish:

Also, if you are using it [the supernova] to measure the expansion of the universe, it is more like a tool. But it is an object as well.

Goldhaber:

That is right. It is an object that is a tool. [laughs] An object that we use as a tool.

Pavlish:

Yes, exactly. The really tantalizing question in this, with scientific objects, is the universe as a scientific object. Has it become a scientific object through this work, or through this theory?

Goldhaber:

Yes. There is a lot of work going on in astrophysics and this was one important ingredient.

Pavlish:

And the third group of questions consists of more personal inquiries about your experiences in the Supernova Cosmology Project (SCP). Some of this is documented elsewhere, so I will not re-ask you how you got involved in the SCP unless you would like to do that. There will be some mixing of question types and perhaps some interjections during our conversation; I would like to leave open the possibility for you to take the conversation in other directions. Right now, why don’t we start with some facts?

Goldhaber:

Yes.

Pavlish:

In your opinion, what are the milestones or turning points in The Supernova Cosmology Project, since you have been with it since very close to its inception?

Goldhaber:

Yes. Well, the first turning point was when, after three years, we discovered our first supernova. We were of course encouraged that there was a discovery by another group, who saw one supernova at a redshift of about 0.3. This came up after we started. It is not that this was our starting point. After we started, this came up, and it demonstrated how difficult it was to find such distant supernovae. Then, we finally, thanks to Saul’s new method, managed to find really sizeable numbers of supernovae. The change was that we got time on a better telescope, a bigger telescope in Cerro Tololo and with that telescope we were able to find something like ten to fifteen supernovae in one night. The method was then to take reference images, and then three weeks later, to take, hopefully, discovery images. That really worked, and it also was then possible to arrange for getting spectra taken. Once we found a good candidate, we also arranged for time at the 10-meter KECK telescope to get spectra taken. That really worked — then we really began to collect a sizeable number of supernovae.

Pavlish:

You had these observing runs, right? Every year you had a certain number of observing runs?

Goldhaber:

Yes. We got about one sequence per semester. That involved first taking references and then looking for the supernovae in new images, and so forth.

Pavlish:

When you take a reference, do you look for the supernovae the night after the reference?

Goldhaber:

No. We look for the supernovae three weeks after, which is about the time it takes for the supernovae to rise. In the supernova rest system, it takes about seventeen or eighteen days, but then with a redshift of a half, you multiply this by one and a half, so it really works out to about three weeks. So, that method worked. And then we had worked out methods of how to do K-corrections, how to correct to the rest system. For my own part, I started then to make a table of these supernovae.

Pavlish:

May we pause, I am sorry, this is a scientific clarification, but I think it is important for me to understand. Here is the supernova light-curve, and what you do is take points along the curve, right?

Goldhaber:

Yes.

Pavlish:

One of the method innovations of the group, or of Saul Perlmutter’s, was to start to look for the supernova here, on the rise.

Goldhaber:

Try to find it on the rise, yes.

Pavlish:

So you point your telescope at the sky and take data.

Goldhaber:

Before it starts…

Pavlish:

Then you wait three weeks.

Goldhaber:

Yes. At first there is no supernova, hopefully there is no supernova.

Pavlish:

And three weeks later you catch it on the rise.

Goldhaber:

We catch it on the rise, yes. We found some quite early.

Pavlish:

Then you follow it.

Goldhaber:

We follow it for about another two months or so.

Pavlish:

Wow.

Goldhaber:

Yes. I mean on an occasional basis. Not every night.

Pavlish:

That is the limitation of the observing time.

Goldhaber:

That is again the question of getting observing time.

Pavlish:

The rise here, you said was 17 or 18 days in its own frame.

Goldhaber:

Yes. In the rest frame but longer in the laboratory frame.

Pavlish:

This is three weeks and this is 17 or 18 days times…

Goldhaber:

Times one plus z.

Pavlish:

Okay, and then you follow it for about two months to get the entire light curve.

Goldhaber:

That is right. We also sometimes get a point after a year, after it is gone. We can look at some light curves and see.

Pavlish:

That was how you found asteroids for example?

Goldhaber:

That was earlier. Before we found any supernovae, I found asteroids. To go into more detail, we actually take two images. We take two images each time. That has three purposes. You need two images to avoid hot pixels and cosmic rays, so that you can rule them out. But also, if you see asteroids, which do show just like a supernova, as new light, then we were able to put a time interval between the two images, like 15 minutes. That is enough time for the asteroid to move. And so what you see then is a new object but it has moved. Then you know it is an asteroid.

Pavlish:

So that is when you are actually doing the search, you always take two images.

Goldhaber:

You take two each time, yes. Each time you take two images. For these three purposes I mentioned. Before we found any supernovae, I found some asteroids. We used to have two or three minutes between the two images. We took the two images one after the other. And then we realized that we needed a little time gap in between to see the motion of the asteroids easily. They can fool you sometimes, but mostly one can observe them moving.

Pavlish:

When I see these light curves, the dots are arranged in different places on the light curve. That is completely due to what telescope is available?

Goldhaber:

That is what telescope is available.

Pavlish:

You always like to get it at peak, right?

Goldhaber:

Yes, we like to cover the peak, to have a point before the peak.

Pavlish:

Before the peak, okay. And the spectrum you like to take at the peak?

Goldhaber:

Yes.

Pavlish:

Okay. Before I interrupted, you were giving the milestones or the turning points.

Goldhaber:

As I see it, I was making these tables. By the way, have you gotten in touch with Rob Knop?

Pavlish:

He is at Vanderbilt, right?

Goldhaber:

Yes.

Pavlish:

Unfortunately this is a limited grant and I was not able to take a separate trip there.

Goldhaber:

Okay. He was doing the measurements on the supernovae. Basically, he drew a light curve by hand and then I fitted it with this SN Minuit program we have, which was written by Don Groom. He is in our group; he is here. You could interview him. So, I was collecting them one or two at a time until finally, I suddenly realized, that there was an enormous peak in the distribution.

Pavlish:

You were collecting the peak brightness?

Goldhaber:

Yes, the peak brightness for each supernova, from a fit to the entire curve. And when I plotted them, I got a very strong signal. At least, to my eyes it was very strong. Other people did not necessarily feel it was significant. But, with my experience, I knew that it was significant.

Pavlish:

What exactly did you plot? You gathered the peak brightness of all the supernovae before you had the spectra?

Goldhaber:

Well, we also got the spectra. We knew they were type Ia supernovae. And we plotted this against the redshift. That is the Hubble plot. I happen to have some here if you want to refresh your memory.

Pavlish:

I guess what is important is accounting for all these possible systematics.

Goldhaber:

I corrected for the stretch. We already knew (this was some work already done by Philips at Cerro Tololo and Hamuy) that there was a correlation between the width of the distribution and the brightness. After you do your measurement, if you also measure the width before you put tem on a curve, you correct for the stretch. If it has a large stretch, you increase the magnitude. If there is a small stretch, you decrease the magnitude.

Pavlish:

You did that for the supernovae that your team had found up to that point?

Goldhaber:

Yes.

Pavlish:

This was the fall of 1997?

Goldhaber:

Yes, it was the fall of 1997.

Pavlish:

Do you have a document showing exactly when you were doing this?

Goldhaber:

Yes, I can give you one. It is unpublished. This is my description of that discovery. [shows document] This is this peak I talked about. I showed this in Santa Barbara on December 14th, 1997. Now there is some contention: who did it first? Each group says, “We did it first.” [laughs] This here is a modern curve. This is with the same coordinates and this is now done with hundreds of supernovae. And there we had 40 at this point.

Pavlish:

This was a conference at Santa Barbara?

Goldhaber:

No, I was invited to give a talk. I showed this.

Pavlish:

This is in Robert Kirshner’s book, “The Extravagant Universe,” right?

Goldhaber:

It is in Kirshner’s book, where he says I did not say anything. I disagree with him. As you will see in here, also, if you read it, I disagree with Kirshner. I showed this and he either did not understand or I do not know what.

Pavlish:

Please explain to me what this is. This is a Hubble Plot? And how do you get this?

Goldhaber:

I get this by making a series of Hubble plots. These points are individual supernovae, where I plotted the magnitude after this correction for stretch.

Pavlish:

This is Figure 1 in your article entitled, “The Discovery of the Evidence for the Acceleration of the Expansion of the Universe: A Personal Account.”

Goldhaber:

Yes. I plotted the magnitude against the logarithm of the redshift. As you can see, it is a mess. But then, I plotted Hubble curves for different values of Omega_Mass: zero, 0.2, 0.4, and so on. Then, I simply counted how many of these points fall in these intervals. And to my surprise, I found that there were two intervals which had eight and ten respectively. Here is a plot of this.

Pavlish:

To get a first estimate on Omega_Matter?

Goldhaber:

Yes. And it is even correct. It is a little more than 0.2, which is the best value now. 0.25 is its current value. The other important thing was, our first seven supernovae are these shaded ones here.

Pavlish:

This is Figure 2 now.

Goldhaber:

Figure 2, yes. And they fell outside of this peak. That had us fooled at the time, which Omega_Mass is close to one. One of these was dropped out — it was not a type 1a we learned later, and some have been re-measured, but basically they are all in this region. Now, why is it that we got seven that were completely outside the peak is not clear. It could be a statistical fluctuation.

Pavlish:

This is dated September 24th, 1997. So this is after the seven supernovae paper, right?

Goldhaber:

Oh yes, right.

Pavlish:

So you did this analysis on your own and you found that your results of your paper were different from what you were getting with more supernova data.

Goldhaber:

That is right.

Pavlish:

Were you thinking about Omega_Lambda?

Goldhaber:

This was under the assumption of a flat universe, which means Omega_Mass and Omega_Lambda equal to one, for a flat universe. So this implied that there is an Omega_Lambda and that there is acceleration. The alternative is actually shown here on this plot. I can probably give you a copy of that. This is what I showed in Santa Barbara. Here is Omega_Mass, plotted. It is somewhere around 0.2. But here, if you assume that Omega_Lambda is zero, you get a different Omega_Mass for this data, a different set of curves. These are all negative, but negative Omega_Mass is unphysical. I knew that something would be wrong if I went with the Lambda equals zero assumption.

Pavlish:

I do not want to skip around too much, but maybe this is an appropriate time for more philosophical questions.

Goldhaber:

Yes.

Pavlish:

I think one of the questions motivating the historical investigation is the title of my advisor Peter Galison’s book, called “How Experiments End.”

Goldhaber:

Yes. I am in his book, “Image and Logic” for my particle physics work.

Pavlish:

Yes. “How Experiments End” has three central chapters. The book consists of three case studies. One is on Einstein’s experiment to determine the ratio between the gyromagnetic ratio and electric currents in magnets. Another case study is on the discovery of the muon — whether it was required to have a new particle or if a revision of quantum mechanics was necessary. There were a series of experiments by people like J. Curry Street.

Goldhaber:

By whom?

Pavlish:

J. Curry Street. He is often attributed with discovering the muon, but there were many other people.

Goldhaber:

In my book, I quote both Anderson and Street.

Pavlish:

The case study is an investigation but it is not a judgment. In his third case study he details the discovery of neutral currents. His goal is philosophical, really. What I would like to get at is how do you decide, say, that the universe is accelerating? There is the personal point, when you believe the evidence, when you are convinced that this is the case. Then, there is also the process, when you are working on a team, you have to convince the others or they have to convince you. When does the team believe the evidence? Beyond that, when does the community believe — either the community of astrophysicists or cosmologists or particle physicists, and even beyond that.

Goldhaber:

Yes. These are three separate steps, I agree.

Pavlish:

Galison writes, that “Experiments end in the Gestalt picture …"

Goldhaber:

The what?

Pavlish:

Gestalt. You know, when you are shown a picture and you either see it as a duck or a rabbit; the picture can be both, but one cannot distinguish the two images at the same time.

Goldhaber:

Gestalt, yes.

Pavlish:

Thomas Kuhn really liked this as an example of how science works, in his “The Structure of Scientific Revolutions.” Galison is arguing against these people. He writes, “Experiments end in that [Gestalt] picture when they fit into a preconceived theoretical scheme. Experiments end in the social construction picture when they mesh with the interests of the dominant theorists.” For Galison neither of these accounts is satisfactory. And so he researches the history of physics to discover how experiments end. I would like to ask, especially in relation to this work of yours… You were finding this result, but neither group published this result right away. You did not announce it right away. When do experiments, or observations, or discoveries, come to their conclusions? How does consensus form around a scientific finding? I suggest that an experiment ends when everyone in the research group agrees with the result, or when it gains general acceptance in the scientific community. But you still need room for individual aha moments.

Goldhaber:

Yes. The steps were like this. I made this sketch, and I was convinced. Then, when I showed it to my colleagues there was a lot of skepticism. Then, they started working on it, using their own methods. Nobody used my method again. They used their own methods to check what Omega_Mass is for a flat universe. And they came up with the same conclusion, because that is what the data showed. I used an unusual method and so not everybody was convinced.

Pavlish:

What about the method is so different? And, did your method finally make it into the paper?

Goldhaber:

No.

Pavlish:

But you were one of the senior members of the group.

Goldhaber:

What we published finally was a distribution of Omega_Mass against Omega_Lambda. And, you get sort-of an ellipsoid; several ellipsoids depending on how many sigma. That was clearly a more conservative method. There, you do not assume that the universe is flat. That was not really known when we did it. It is later on that the CMB measurements showed that the universe is flat. It was an assumption on my part based on the theory of inflation. Inflation assumed that the universe is flat and so I just made that assumption. But that was not necessarily accepted. So what we finally published is a probability distribution. And also, many more corrections and checks were made. In other words, I did the simplest possible thing and then my colleagues made many more checks.

Pavlish:

Did anyone else go back and do the simplest possible thing?

Goldhaber:

Not to my knowledge, no. [laughs] Only I am still doing that. If you look at what the significance is of this distribution, the significance is that there are different universes, which have different values of Omega_Mass. If you read it literally, that does not make any sense.

Pavlish:

The results show that there could be different universes with different Omega_Mass, right?

Goldhaber:

I cannot rule that out, yes. It is consistent with that, but what it really is, is one Omega_Mass and experimental error. So, this method has not caught on. Nobody uses it except me.

Pavlish:

Is this method based on methods you used in particle physics?

Goldhaber:

Yes. In particle physics I always looked for peaks. I found charmed mesons by looking at a peak. I found the A-meson and many such things. Other people found a lot of different states. You look at a distribution and a histogram and you find a peak in the histogram. That is indicative of some effects taking place.

Pavlish:

What would you usually plot on the axes in your particle physics work?

Goldhaber:

This axis would be the same: number of events. But here would be the mass of a state. Say, for instance, you take two mesons, a K and a pi, K-pi mass, and you plot it and then you get a peak where there is a charmed meson, which has decayed into the K-pi. So, in other words, you plot the decay products, the mass of the decay products, and you find a peak where there is a resonance.

Pavlish:

You did that for multiple particle physics experiments?

Goldhaber:

For many different experiments, yes. But, I did it particularly for the K-meson in that sense of looking at a histogram and finding a peak Now, Saul presented this first result at a talk in Santa Cruz, already in (I do not know the exact date) October of 1997, probably. The point is coming up now with the other group: who had something first? Well, I gave this talk on December 14th in Santa Barbara, but Saul gave a talk even earlier in Santa Cruz.

Pavlish:

Which talk did you give in December?

Goldhaber:

The talk about sharing this result. The talk included other things, but I showed this result.

Pavlish:

In Santa Barbara?

Goldhaber:

In Santa Barbara.

Pavlish:

Robert Kirshner was there?

Goldhaber:

Kirshner was there. But he somehow then, later, in his book, he did not acknowledge that I showed it.

Pavlish:

Was anyone else from their team there?

Goldhaber:

Yes. They are all mentioned in this paper — the whole list of people. In particular, David Gross said, “Well, can you be sure with such small statistics?” I said, “Yes, I am sure.”

Pavlish:

How about within your team?

Goldhaber:

I presented this picture that is in here.

Pavlish:

You have written it up?

Goldhaber:

Yes.

Pavlish:

Then, maybe we can get on to the other questions.

Goldhaber:

Okay, sure.

Pavlish:

In terms of milestones then for you personally, you were convinced by this.

Goldhaber:

Now, there was some other evidence in the air. For other people they may find those more convincing.

Pavlish:

What kinds of evidence?

Goldhaber:

We had actually just published a paper in Nature. Because of one event, it changed the original seven where the Omega_Mass was 0.9 something to 0.6 something.

Pavlish:

That was in January, though, right?

Goldhaber:

The other paper was sent in within a few weeks of this. Then, the other group had three supernovae which also indicated an Omega_Mass of 0.2. My point is, since we had shown that seven supernovae can give you a completely wrong result, you have to be careful of the statistics. You cannot just say, “Okay, I have three and that is it.” My claim is that this was the first real proof. But everybody will give you a different answer as to what convinced them.

Pavlish:

Did you manage to convince anyone with this plot?

Goldhaber:

Oh, yes. Rob Knop was convinced. That is why I wanted you to talk to him. Here he is saying that indeed, I was the one that found this. Here is a copy, this is for you.

Pavlish:

Thank you.

Goldhaber:

It says this is preliminary. I showed it just like this.

Pavlish:

So, Rob Knop was convinced by this analysis.

Goldhaber:

Rob Knop was convinced.

Pavlish:

How about Saul Perlmutter?

Goldhaber:

Well, he was convinced enough to show it in his talk in Santa Cruz. So, he certainly was strongly influenced by it. You see, the data was there. It was a question of looking at the data in whichever way you wanted to. And, they got more convinced by looking at it themselves.

Pavlish:

Is that typical of collaborative experiments, do you think, that each member wants to look at the data in his or her own way?

Goldhaber:

It happens to be in this case. It is not necessary. It depends how good the data looks. When we found the psi, there was no question. Nobody doubted it, because it was a very clear observation. (I gave you some literature on that, on the psi). Here it had several strikes against it. The first strike is that our first paper did not see it. Secondly, I used an unusual method. Thirdly, nobody wanted to believe that the universe is accelerating. In other words, the result was unbelievable, so to speak. So, with three strikes, people had to convince themselves. I would not say that it is always the case that everybody has to repeat all the calculations, but in this case it was necessary.

Pavlish:

Had you done this method also for the seven supernovae paper?

Goldhaber:

No, I had not done it then.

Pavlish:

So what made you decide to use the method at this point?

Goldhaber:

I cannot say. It just occurred to me, that here is a way I can look at it.

Pavlish:

Were you working in your office here?

Goldhaber:

Yes, I was working in my office. And I produced this graph on the computer and then I just went and counted, and wrote the numbers up here. There were 38 supernovae at that time. Later on, we had 42, but when I first did this, there were 38.

Pavlish:

Had that data just come in?

Goldhaber:

No, it came in by bits and pieces. It took quite a while to analyze the supernovae. We got one or two coming in per week, which we had already measured. I kept tabs. Anybody was able to do that and I did it.

Pavlish:

So you have tables?

Goldhaber:

Yes, I can also give you copies of those tables.

Pavlish:

Why did the supernovae come in one or two a week?

Goldhaber:

The measurements got completed as we took more points along the curve, and we were then able to fit the curve. With the first point we discovered the supernova, but we could not yet fit the curve. We had to wait for the data to come in, for the curve.

Pavlish:

So you had observing runs where you would find a certain number of supernovae?

Goldhaber:

We had observing where we would find the supernovae, but then we had runs where we followed the supernovae. It sometimes was difficult. It involved more than one telescope.

Pavlish:

Like you said, you followed it over months?

Goldhaber:

Yes. Two months up to a year. Sometimes we needed to get a final point in order to subtract the effect of the galaxy. You see, the supernova can be near the edge of the galaxy, but if it is near the center, then you have to subtract the galaxy. For that you have to wait for the supernova to die out. So, it was a drawn out process. It was not a question that you find it and you have a result. The situation is that you find it and you start measurements. So, in that sense it is much different from the usual, from what happens in particle physics.

Pavlish:

So there is no analogy in particle physics where you would find something and then continue measurement?

Goldhaber:

Yes.

Pavlish:

You find a cosmic ray, find an event, and it is done.

Goldhaber:

Well, you have to still analyze it, but it happens fast as opposed to this.

Pavlish:

Any other milestones or turning points, do you think, in the early history through 1999 that you think are important? Did you pay attention at all to the “Great Debate” of 1998 between Peebles and Turner in The Smithsonian in Washington, D.C., entitled “Is Cosmology Solved?”

Goldhaber:

I do not recall the date.

Pavlish:

It was in the fall of 1998. Then, in the spring of 1998 there was a Fermilab grilling, a call for members of the two teams to defend your results to the greater community and they took a poll at the end about how many audience members believed the result. I wonder if there are any events that stand out in your mind towards understanding how the greater community came to accept the result.

Goldhaber:

Not offhand. It was debated back and forth a great deal because it was such an unusual result. Nobody wanted to believe it.

Pavlish:

How has the SCP changed over time?

Goldhaber:

We had people coming and going. A few basic people have remained for quite a while. Rob Knop is one of the early senior people who left. But, Peter Nugent, Greg Aldering… [end of tape side 1 cuts of this list] [tape side two]

Goldhaber:

We developed a CCD here at Berkeley. And also there were our collaborators, Reynald Pain (he is actually visiting here now) and Ariel Goobar. They have remained collaborators and are still active in the same field, although they are now also working with other people. For a while we were also working with Filippenko. When the other group formed, he decided that he was happier working with astronomers than with this bunch here. He joined the other group. Now that we have won the Gruber prize, I think that he gets it double. [laughs]

Pavlish:

[laughs] That is funny. I have seen a picture of you, Schmidt, and Reese, the three of you sitting together in Aiguablava, Spain. Adam Reese showed it to me, and I hear that it is one of the few times that you were together socially. It is a nice photograph. Do you remember anything specific from the conference?

Goldhaber:

I remember being there. I do not remember that photograph. I do not have a copy of it. I sent Adam Reese copies of the photographs that we made, but he never sent me that one.

Pavlish:

You took photos there too?

Goldhaber:

Yes. Jude and I were there. Somehow we took some photos.

Pavlish:

Of the members of the two teams?

Goldhaber:

There was a place where they had Hebrew astronomers and I took a picture of Adam together with that somehow. I do not remember the details.

Pavlish:

You gave the milestones for you and within the team, but how about in the community of people doing observational cosmology? Were there important meetings where you exchanged significant information?

Goldhaber:

I did not personally go to all of them. Saul went, mainly, to those meetings.

Pavlish:

That was one of the advantages of this group, that you had the center of many people here working on the project.

Goldhaber:

Yes.

Pavlish:

You did not necessarily have to go elsewhere.

Goldhaber:

Not everybody went to those conferences.

Pavlish:

In terms of the division of labor in your group, did you do this work on your own or did somebody ask you to do it?

Goldhaber:

Nobody asked me to do it. As a senior member of the group, I decided that this was a way to look at the data. First of all, just to collect the data, it started coming in very slowly. Here there were 38 supernovae, and then when we finally published, there were 42. In fact, there were 38 in this paper. In this note I just gave you, there are 40. It kept increasing.

Pavlish:

Okay, so in September of 1997 there were 38?

Goldhaber:

Yes, and this is December.

Pavlish:

If I were to ask, when did your team first think that there could be a nonzero cosmological constant, would it have to be different for every team member?

Goldhaber:

Probably. Some felt that there was some evidence when we added an eight supernova to the seven, with the Nature paper. By the way, that went in after I found my effect, though the data was being analyzed earlier.

Pavlish:

But your findings did not really affect the paper?

Goldhaber:

No. I do not think it affected it, no.

Pavlish:

Another question of mine is, when did you begin to think of Lambda as Dark Energy? I assume that you thought of it as Lambda… until this point.

Goldhaber:

We thought of it as Lambda, as Omega_Lambda, and Dark Energy is a name that Michael Turner gave it, to have a nice name.

Pavlish:

After?

Goldhaber:

Of course after, yes. After he convinced himself. By the way he did not believe my result either. I gave him a copy of it and he said, “No, I do not think that is it.”

Pavlish:

He wrote a paper in 1995 entitled something like “The Cosmological Constant is Back.”

Goldhaber:

Yes.

Pavlish:

Were you aware of that? Did that influence you at all? The paper was not based on supernova measurements.

Goldhaber:

No, no. It was based on other measurements. I am aware of that paper now. I am not sure that I was aware of it at the time. Anyway, it did not influence me. I just went by what data I had.

Pavlish:

Mike Turner was on The Review Board or something, for the project?

Goldhaber:

No, Kirshner was. He said, “This will never work.” In 1989-1990 he was on the board. He essentially said, “You cannot make that work” and “Do not give them any money.” Let’s see, here is a short version of the table. Or, do you want exactly the original table?

Pavlish:

Well, if you have the original, whatever is clearer to the historian. Did you use any of the same plotting software that you had used in particle physics?

Goldhaber:

No. I had to switch to IDL.

Pavlish:

This is not something I had prepared, but it occurs to me that when you joined the group, you probably had to read up a bit on cosmology.

Goldhaber:

Yes. I still was not an expert yet. So maybe that is why getting this strange result did not phase me so much as it did astronomers. [laughs] I did not know enough. Okay, that is the table in all its glory.

Pavlish:

Do you continue keeping such a table now?

Goldhaber:

No. I stopped with the first 42. I do not think that anybody has kept up these tables.

Pavlish:

What do you have here? It looks like a lot of numbers.

Goldhaber:

The name, the official name, the redshift, the B-magnitude, the error on the B-magnitude, and then the corrected B-magnitude…

Pavlish:

What does B-magnitude mean?

Goldhaber:

B is the color, blue.

Pavlish:

Okay.

Goldhaber:

Then a different correction, then the error on the correction, then the stretch, the error on the stretch, then the color, error on the color.

Pavlish:

This was already after you had realized that blue was a more trustworthy place to look?

Goldhaber:

Yes. We had color information for some of them but not for all of them. [To continue explaining the table] I forget what this is, then percent increase, then the extinction, then the spectral quality. This was important. It is something I did with Richard Ellis, to see which supernovae were really well established, as opposed to only likely. Then, the K-correction, another K-correction, then the Chi-squared degree of freedom, the Julian date, and then there was some covariant information which finally Greg Aldering added some three items to the table involving covariance.

Pavlish:

Then you crunched those numbers to get the maximum of the light curve, to plot against the redshift?

Goldhaber:

This is the maximum.

Pavlish:

It does have the maximum in different colors?

Goldhaber:

This is the maximum. What I call “MB” is the maximum.

Pavlish:

Maximum Brightness?

Goldhaber:

After correction for stretch.

Pavlish:

Did you do that for different colors?

Goldhaber:

No. We only did that for B. We measured in R, with an R-filter, and it came out then, in B.

Pavlish:

Where is the redshift on this table here?

Goldhaber:

Redshift is here: z. It is the third item.

Pavlish:

You do not include the nicknames of the supernovae on here?

Goldhaber:

Yes, we do. We have our name and then the official name. “94F” is an official name, 1994F.

Pavlish:

Where is your name?

Goldhaber:

Just before that.

Pavlish:

That is a number too?

Goldhaber:

It is a number. The first two digits are the date. 94 is the year. And this is the number that we gave it when we were finding it. The numbers are quite large because we found a lot of junk and the supernovae were among the junk.

Pavlish:

Later you also started naming them after classical music composers?

Goldhaber:

Oh, yes. Then we named them after composers. Both Saul and Rob Knop are musicians, so they chose composers. Recently we have named them after family members. We have a whole set now named after family members. I can give you a list of those family names.

Pavlish:

Thank you, but I think that would be difficult to justify as historical material, if it is happening right now. Even this history of the 1998 discovery is a little bit on the border of history.

Goldhaber:

Yes, this is what is happening right now. Back then we only had numbers. We did not have special names. Later on, after the first 42, we went over to composers.

Pavlish:

That was after the 42, okay, I see.

Goldhaber:

Yes.

Pavlish:

Did you use other kinds of visualization techniques for thinking about the data; aside form the one you have shown me here? Like graphs, for example?

Goldhaber:

Yes, innumerable graphs.

Pavlish:

Innumerable?

Goldhaber:

Innumerable! [laughs]

Pavlish:

Of what?

Goldhaber:

Well, first of all, all the light-curves.

Pavlish:

But those are predetermined, aren’t they?

Goldhaber:

What do you mean?

Pavlish:

You just plot the data?

Goldhaber:

Yes.

Pavlish:

Is there a lot of data reduction that goes into developing the light-curves? I ask, because when developing this theory of scientific objects, what people think about is the relationship between the representation of an object and the object itself.

Goldhaber:

Right. I am picking a random set, from December 1997. Here is how we find a supernova. This is the image.

Pavlish:

Oh, look at those. Wow.

Goldhaber:

We have all kinds of representations.

Pavlish:

This is all done with one computer program?

Goldhaber:

Yes. This is a computer program which we used for scanning.

Pavlish:

What are these? Topologies?

Goldhaber:

Yes, topologies. The thing is, the program found supernova candidates, but it found ten times more stuff. So then, we needed human intervention to see which ones were really supernovae.

Pavlish:

Was that true in particle physics too? That your computer would find candidates and then you would have to find the real particles?

Goldhaber:

No.

Pavlish:

Rather, the person would be the computer in particle physics.

Goldhaber:

Here you have a nice example. This was the galaxy before. And then, suddenly it grew a little extension. This is then subtracted. You subtract this one from that one and what is left is the supernova.

Pavlish:

Why does the topology have these rings?

Goldhaber:

You see how bright it is. It is brighter towards the center. These are rings of different brightness. Here, then, is the picture of it.

Pavlish:

Of the brightness?

Goldhaber:

Yes.

Pavlish:

So you zoom in on the topology?

Goldhaber:

Yes, right.

Pavlish:

Wow. Now what are these numbers?

Goldhaber:

This is the numerical description of that peak.

Pavlish:

The computer program would find more?

Goldhaber:

Yes, you see here I have all of them and occasionally we say it is a supernova. I guess I used green for supernovae. I did a lot of work on the scanning.

Pavlish:

You printed it out and then you labeled them. Green post-its for supernovae. Purple post-its for junk.

Goldhaber:

Yes. Here is another case.

Pavlish:

Wow. These are all dated, too.

Goldhaber:

Yes.

Pavlish:

And then you give your analysis, like “Could be a nice bright supernova with a lot of light in the reference or it could also be an AGN.”

Goldhaber:

AGN is a galaxy which changes in brightness because of some black hole interaction that suddenly emits light and then goes up and down. Or the black hole swallows a star and emits a lot of light. That is always then in the center of the galaxy. The supernova…

Pavlish:

Can be anywhere in the galaxy?

Goldhaber:

Yes.

Pavlish:

Can you tell from these images whether it is an elliptical or a spiral galaxy?

Goldhaber:

Not from these, but by looking at it in more detail, you can.

Pavlish:

What does that mean? You say that it could be a nice bright supernova?

Goldhaber:

There is a comment here, by Rob. (The scanner’s name is also given.)

Pavlish:

Oh, wow.

Goldhaber:

“It could be a nice bright supernova with a lot of light in the reference. Or, it could be also an AGN, which is part of a pair of interacting galaxies. Further subtractions are warranted.” So he was not sure. But then, later on, I was sure. This is December 28th, 1997, so this is post-42. I can find one which contains the 42.

Pavlish:

Oh, you mean this is even further data?

Goldhaber:

This is later, yes.

Pavlish:

But you were doing it in 1997.

Goldhaber:

Yes, but it was not ready for publication.

Pavlish:

When did you stop taking data for the 42 paper?

Goldhaber:

Early 1997.

Pavlish:

Early 1997, okay.

Goldhaber:

Yes.

Pavlish:

Did your data over the years 1997 and 1998 affect that paper?

Goldhaber:

No. We had had more supernovae but they were not fully analyzed, so we could not use them. You have to essentially wait a year. Here, we had them in 1997. There are, in here, oh quite a few from 1997.

Pavlish:

But those were not included in your 42 paper?

Goldhaber:

Let me think a moment.

Pavlish:

That would be interesting to know: when did you stop collecting those 42. Although, maybe the date of each supernova is in the paper itself? I can check it.

Goldhaber:

The paper has only the official names. So, let’s see. This is December. When I did my analysis it was September of 1997. So, this is afterwards. This is post that paper.

Pavlish:

Do you have a binder like this for the previous data?

Goldhaber:

Yes.

Pavlish:

Were these helpful also in your visualizations?

Goldhaber:

Moderately helpful. This table was more helpful.

Pavlish:

The numbers were more helpful?

Goldhaber:

The numbers were more helpful. See, not everyone wrote a whole story. This guy said, “AGN?” Start, December 1997. Let me see what this is. “Good supernova candidate.” “Needs more study.” “Bad.”

Pavlish:

Green, pink, and purple are the code colors.

Goldhaber:

It is coded by color, yes.

Pavlish:

Green is a “good supernova candidate.” Pink is what?

Goldhaber:

Pink is “needs more study.”

Pavlish:

Purple is bad. Wow.

Goldhaber:

That is my nomenclature, yes.

Pavlish:

It looks like you have a lot of green here, a bit of pink, and a bit of purple.

Goldhaber:

Let’s see. 190.

Pavlish:

December 29th. That would be beyond. Well, it could have been one of the last ones, right?

Goldhaber:

No, it cannot, because this is the discovery, and it takes a whole year to measure the light curve.

Pavlish:

Oh, so you search and you find.

Goldhaber:

Yes, this is just finding it.

Pavlish:

This is just finding it. Oh, wow. And then you decide to follow it up or not.

Goldhaber:

Yes, right. This is another thing. I am the one who kept all these records. I still liked paper and everybody else just works on a computer. Here is what some of these light curves looked like. Let’s see. Here is a light curve of 95103. And you see that we took data a year later to really anchor it down.

Pavlish:

So this is a notebook of light curves. That would be a year delay. After you get this notebook you have a similar notebook of light curves.

Goldhaber:

Yes. At some point I also gave up on keeping paper records.

Pavlish:

How about before this. Do you have a notebook like this for the data of the 1999 paper?

Goldhaber:

I think so.

Pavlish:

Oh, these are 1995, 1996.

Goldhaber:

Those are the ones that are in the 42 paper.

Pavlish:

You have a notebook that says on the cover, 1995 and 1996 supernovae, R and I joint-fit. What does that mean, R and I?

Goldhaber:

We had two colors. The filters are called R and I.