[Bernard Mills]
Bernard Mills, 2006 (Photo courtesy Chrys Mills)



NATIONAL RADIO ASTRONOMY OBSERVATORY ARCHIVES

Papers of Woodruff T. Sullivan III: Tapes Series

Interview with Bernard Y. Mills
At Grenoble
26 August 1976
Interview time: 165 minutes (parts 1 and 2 combined)
Originally transcribed as typescript only by Bonnie Jacobs (1977), retyped to digitize by Candice Waller (2016)

Note: The interview listed below was originally transcribed as part of Sullivan's research for his book, Cosmic Noise: A History of Early Radio Astronomy (Cambridge University Press, 2009). The original transcription was retyped to digitize in 2016, then reviewed, edited/corrected, and posted to the Web in 2016 by Ellen N. Bouton. Places where we are uncertain about what was said are indicated with parentheses and question mark (?).

We are grateful for the 2011 Herbert C. Pollock Award from Dudley Observatory which funded digitization of the original cassette tapes, and for a 2012 grant from American Institute of Physics, Center for the History of Physics, which funded the work of posting these interviews to the Web. 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.

Click start to listen to the audio of the continuation of tape 62A of the 1976 interview.

Begin Part 2 of Tape 62A

Sullivan

This is continuing with Mills on 26 August 1976. So we’re looking at your first large survey in 1952, in which you define the Class 1 and Class 2 sources and you were just telling me about this paragraph on page 271.

Mills

Yes, where I had stated that for the purpose of compiling a list of sources and for subsequent analysis, a discrete source has been defined in terms of a particular kind of pattern produced on the pen recorder, etc. This indicates the sort of operational definition I’d been forced to adopt because it was clear that we had different patterns on our two interferometers. The wide-spaced interferometer in general didn’t always have the same pattern as the close spaced one. So obviously you were not dealing with a single point source all the time. So one had to make an operational definition. This, I think, is a source of a lot of the disagreement between Cambridge and us. That they were not forced to a sort of definition by working at one spacing only, they thought in terms of single discrete sources. The other thing, of course, was the repeating patterns we got, which indicated resolution as a problem. So that, again, one separated this out in terms of periodic patterns on the recorder and you defined each pattern as a discrete source. But I did want to make it quite clear, in here, and may not have used quite sufficient emphasis – that it was purely an operational definition.

Sullivan

Right, I can see that now that you point that out. Looking towards the end of this same paper you have a log-n log-s in the middle of the paper. Then at the end, you talk about whether they really are extragalactic they would be of great importance to cosmological theory. And you were telling me before about how you spent a lot of time. Could you tell me about that again?

Mills

Well, I took a quick glance through to see just what I did say. It was a long time ago. But, I think the point was we had a whole set of different observations. We had these isotropic weak sources, concentrated strong sources, and we had some possibility of identifications with extragalactic objects. We had a possibility of identifications with galactic objects. These set a series of constraints on the sort of theory which you could have. It took me a very long time to sort this all out in my own mind as to what was permitted by the logic of the whole system.

Sullivan

And you said you actually set up a logical flow diagram?

Mills

Yes, I did. I had to use this at one stage.

Sullivan

You think there’d be any hope of digging that out? That would be of great interest to see that.

Mills

I don’t think so. I have a habit of – when I was thinking about things like this – to scribble them on bits of paper. It could possibly be in one of my old notebooks. I doubt it, though. I probably just threw it away.

Sullivan

But anyway, the result of all this thinking was with what, would you say, a sort of leaning towards the idea that they’re probably - ?

Mills

I think I felt at that stage that they were pretty certainly extragalactic objects – galaxies, in fact – there was no mention or thought of quasars at that time, of course. I felt I had to check this, and this lay behind the angular size interferometer, the radio link interferometer – this was a way of checking, really, whether they were all extended like galaxies.

Sullivan

Now going on to this one, Mills ’52, in which you have six sources with very accurate positions. I’m just looking at specific questions I may have scribbled here sometime in the past. You refer to calibration and refraction effects. Now when you say refraction, do you mean scintillation also, or do you just mean what we today would call tropospheric refraction or is it tropospheric and ionospheric?

Mills

I was thinking definitely in terms of ionospheric refraction. In those days, at low frequency, you could practically ignore the tropospheric effects.

Sullivan

Ok, and scintillation, I think you used the term “random refraction” here. Well, you say ‘the random errors due to refraction effects – that’s what threw me. I normally think of that as a systematic error.

Mills

No, I was thinking in terms of the results we had obtained before, I’m sure now, looking back on it – our Cygnus results. We had obtained apparently slow variations in amplitude which we had attributed to refraction effects. It was probably this I was thinking of here, that these could give apparently systematic changes in the amplitude of the interference pattern. And it wasn’t clear to me that this may not be worse at wide spacings, and this would depend on the model one had, I think, of the ionospheric irregularities.

Sullivan

So it really was a sort of scintillation. You call it random refraction because the source was jumping around.

Mills

Yes, I’m pretty sure that’s what I meant.

Sullivan

And here I’ve noted that you have corrected the position of Cygnus from Mills and Thomas, which was 3-1/2 seconds of time off. Apparently, sometime in the meanwhile you -

Mills

Oh, this is the position of the actual galaxy.

Sullivan

No, slightly wrong, this is the radio positon.

Mills

Oh, yes. The nebula was slightly wrong.

Sullivan

For Fornax A, it turns out that the right ascension is a good bit off and I just want to ask you if you can remember why that was? It turned out to be 55 seconds off.

Mills

Oh yes, in fact, this prevented us making the identification of [NGC] 1316, although we knew the possibility of that identification for quite some time. In fact, I remember Bolton had noted a similarity between that and [NGC] 5128, which he’d read of in on of Shapley’s papers. So we knew about the possibility of an identification with [NGC] 1316. This RA was rather surprising because I was thinking that it would turn out as an identification. The reason is quite clear when you look at the distribution – this demonstrates the difficulty of dealing with interferometers when you have an extended source. Fornax A is a double source and although the centroid is nicely on the galaxy, in fact the fine structure is more distributed on one side than the other. So that the position is weighted towards the position of the fine structure. We know now what’s happened, but at that time it wasn’t clear.

Sullivan

You also mentioned in the paper that this was the weakest source investigated, which didn’t help either.

Mills

No, that’s right. The pattern was not clear. There was a bit of confusion in the pattern.

Sullivan

Well, let’s take a look at this one [1953] AJP on four extended sources. I guess we haven’t talked about this at all. Maybe you can tell me about this experiment in which you made some simple assumptions in order to get the brightness distribution.

Mills

This was an extension of the work, of course, described in that earlier Nature paper, using the same instrument, and merely trying large numbers of different orientations and baselines.

Sullivan

The entire Fourier theory I guess was pretty apparent as to how one would -

Mills

Well, that goes back to Pawsey and Payne-Scott, McCready.

Sullivan

Yeah, that’s right, that paper.

Mills

It was basically Pawsey’s idea. So we were very familiar with that.

Sullivan

Here’s your portable equipment. Was that still Army surplus, or were you buying new stuff by that time?

Mills

We were making a lot of it by then so I can’t really be sure. I don’t think it was Army surplus by then.

Sullivan

Are there reasonable prints of these things still around CSIRO?

Mills

I think it’s probably quite likely there are, yes.

Sullivan

Because I’d like to get hold of some of those. Now, you have a figure here in which you compare photographs with what we’d today call ‘radio-graphs’ of Centaurus, Virgo, and Taurus. Virgo is a little bit confusing, I guess, because its E0 or E1 and you came out with a more elliptical thing.

Mills

This elliptical thing is now known as the halo of Virgo A.

Sullivan

Right. But Centaurus and Taurus really did have basically the shape of the galaxy.

Mills

Which was actually wrong in the case of Centaurus, of course, because again, by having insufficient spacing, I’d missed the double nature of the central source. So that what I came out with was quite the wrong direction. It should have been going at right angles to my distribution.

Sullivan

To the one you had, right. But that was with only three or four spacings that you were trying to make these models?

Mills

Yes. Of course, there was no real indication then that one needed a lot of spacings to sort things out.

Sullivan

Well, here they are here. It looks like for Virgo you have about eight. Let me ask again… there are some interesting wiggles in Centaurus.

Mills

There are, certainly (laughter). Actually what had been missed there was that the visibility function crossed over into the negative region. If I had picked that, it would have made a completely different model. But it did show, grossly speaking, the sizes were equivalent and, in fact, the shapes turned out quite well for the Crab and for M87. Centaurus was missed.

Sullivan

Let me ask more explicitly about this Fourier theory because it’s of interest for me to know whether it was quite clear to you, as you went around to these various places and all, exactly what you were doing. In that if you had had enough time, did you realize that you could have done the whole thing properly?

Mills

Oh, yes. That was known, but it wasn’t considered to be technically feasible.

Sullivan

And why not?

Mills

Because of the phase problem. Simply, the amplitude information was only useful if one had a completely symmetrical system, and we didn’t think it likely that we had that. We felt we had to have phase information if we were really going to plot things. I didn’t think it was technically feasible to carry the phase information at the large spacings and get high resolution.

Sullivan

Did you ever try it, actually?

Mills

Not really, no.

Sullivan

The next thing I come to here is the whole Mills Cross and you mentioned briefly yesterday how the idea came to you. That you really felt that you needed a pencil beam. Could you explicitly say that the Cross and the phase in-phase technique was not obvious at that time, I don’t think.

Mills

No, it came out of our three-element survey. That’s the one we were talking about earlier. There were two things which came out there, I think. One was this difference in the responses of the closed-spaced interferometer compared to the wide-spaced interferometer, which showed we had to have all Fourier components, as we now say. The other thing was we thought that there were resolution problems because we had lots of beating patterns which showed that sources were not being observed individually. I think these were two of the important things. Then the technical thing which suggested to me the use of the Cross was our attempts to sort things out- sources out – when they were beating together by pointing our antennae in different directions so that we were getting the product of the two diagrams. I was familiar with that at the time and it’s mentioned somewhere in the paper that we did this. We used to put out, for instance, a zero on one of the sources and we’d try and -

Sullivan

This was radar technique also, or was it - ?

Mills

No.

Sullivan

Didn’t they used to put an object on the side of beam in order to -

Mills

Yes, that could have been. I don’t remember how it came about that I was using this, but I think I had simply noted that you could cut down the response of a source by putting one of the antennas near a minimum.

Sullivan

Well, that must be this remark which indeed I was going to ask you about. “A similar method of using a common portion of overlapping beams to produce a narrower effective beam has already been used.”

Mills

Yes, that’s right. That was what we were doing. And this really suggested it to me. I think Cambridge were also thinking of this sort of thing at the same time. Obviously, because it came out later in a paper by Ryle.

Sullivan

Well, that was ’52 also. Yours came out in ’53, actually.

Mills

Well, this paper came out in ’52, You see, we had been playing with this sort of technique. And, in fact, the whole idea of the Cross was worked out in ’52.

Sullivan

Did you know about Ryle’s whole phase switching thing while this was going on?

Mills

Yes, we knew of the phase switching technique. This was used in the survey which we had done, in fact, it was mentioned there that I used a phase switch similar to Ryle’s. This was a survey done in 1950 and 1951. So that was a great advance at the time, and we recognized that and we went over entirely to the switch system. And so I felt familiar with how it worked. The things which appeared were that to do surveys you need a pencil beam and I spent a lot of time thinking first of large arrays because it was also obvious that one needed high resolution. And then it did occur to me that you didn’t need all the area of the arrays because you were confusion limited. All those ideas were sort of worked out in ’51, ’52 and I can’t remember exactly when. I can’t remember precisely when the idea of cutting out most of the dipoles in an array and just using a cross came to me. But that’s how, in fact, I thought of it, in terms of an array.

Sullivan

Then it didn’t need to be a filled thing -

Mills

Then I realized that just having a Cross and taking total power was useless, because of the wide responses. And it suddenly came to me, well, I can use the same sort of technique I’ve already been using and just multiply one against the other. That was sometime in ’52 that the idea came to me. So it didn’t really come as a basis of an interferometer, thinking in terms of interferometers, but thinking in terms of pencil beams. I was quite convinced this was what was needed for a survey.

Sullivan

Going back to this comment about using a common portion of overlapping beams. You say that it was used in the survey, Mills ’52, yet I don’t think there’s any mention in that paper.

Mills

Yes, there is. I’m sure you’ll find a mention there somewhere for sorting out beating patterns by pointing antennae in different directions.

Sullivan

Ok, I’ll take a look at that. Well, then you built a model of this. I guess it wasn’t clear that it really was going to work.

Mills

No, there was considerable internal criticism, I’d say. (laughter)

Sullivan

What was the nature of this?

Mills

No, I had to convince the head of our laboratory, Bowen, it was worth putting any money into it.

Sullivan

But why wouldn’t it be?

Mills

I don’t know. He didn’t believe it would work. I can’t, for the life of me, now remember the argument used, but it was something to the effect that if you had two separate antennas, they must give a null result over the sky, so you’d have no – it wouldn’t give any response at all, or something like that. I can’t really understand it. I couldn’t at the time, but I had to get the word to build one.

Sullivan

And so you did and I guess it worked famously.

Mills

Oh, yes, very nicely.

Sullivan

And then came the business of trying to get the full thing built. And I guess you were able to convince Bowen then?

Mills

Oh, there was no doubt about it. After all, we observed the Magellanic Clouds for the first time with that thing – the small one.

Sullivan

Well, since you brought that up, was that a publication?

Mills

No, I think it’s just mentioned as a by-product in this ’53 paper. [Turning pages]

Sullivan

You have a thing of the sun here. Here is one figure showing the observations of the Magellanic Cloud.

End Tape 62A

Click start to listen to the audio for Tape 62B of the 1976 interview.

Begin Tape 62B

Sullivan

Continuing with Mills on 26 August 1972. How much did this large Cross cost, if it’s of interest?

Mills

Oh, very small, around about – perhaps 10,000 pounds or less. But it was constructed almost entirely in the laboratory so it was very difficult to cost – I mean labor, the work force there just worked on it. But the external cost was less than 10,000 pounds. I should have said there was a little more to it than this in building the first experimental one. I had to convince people it would work, but there were also a number of basic problems I wasn’t quite clear about myself which I wanted to experiment with, principally on this matter of the intersection of the two arms, and what was the best way of dealing with it.

Sullivan

How they interact with each other -

Mills

Yes. The experimental model was very useful from that point of view.

Sullivan

And what about the grading? Was that obvious as to how to do that best?

Mills

No. There were several possible ways of doing that. Well, it was just a matter of thinking them out and deciding which in the end would be the easiest to make and to maintain.

Sullivan

Perhaps this is a good place to ask you about how you went about building something like this instrument. Was it a matter that you sat down and designed it, and, more or less, supervised the entire construction of it? Or was it a matter that you were out there with people having fabricated things but you had to be pouring cement and all this sort of thing?

Mills

No, actually I got out of most of the hard physical labor involved, because I went overseas for a while, while it was being constructed. The original design I did almost entirely – the electrical design I did almost entirely alone. The mechanical design, just the physical construction of the thing, was done by McCallister who was our mechanical engineer in the place. I got all the major things sorted out at that time and then Alec Little joined the group. Oh, no, he’d been working before. He did a lot of the test equipment development for it. We were sort of working together on this. He was developing equipment to set it up while I was doing some of the basic design work. Then when we had it all figured out and I had this opportunity to go overseas for about six months, and at this stage most of the bits had been made and the assembly had been starting by our workshop at Fleurs Observatory – it wasn’t an observatory then, just an open field. The actual setting up of it and the adjustment of it was Alec Little’s responsibility and he did a very good job on this. He, as I said, designed all the test equipment which we used for setting up and getting the currents right and the phases right and things like that. We had discussed it all beforehand and we had a pretty good idea of how it was going to work but he was responsible, actually, for getting that all set up. When I came back the thing was almost complete, as far as I remember. Still some adjustments to be made and things like that.

Sullivan

It must have been a pleasant sight?

Mills

Oh, it was a very pleasant sight.

Sullivan

What determined the overall size of the thing? Was it just constraints on funds or was it the size of the field or was it something astrophysical?

Mills

Well, the astrophysical requirements set the minimum size of the thing. It had to be beyond about 1,000 feet, something of that order. The frequency was chosen on the basis of where we thought the maximum sensitivity for point sources would be. In fact, there’s a bit of discussion of that in a later paper in the astronomical issue of the Institute of Electrical [i.e. Radio] Engineers ’58 -

Sullivan

Ah yes, that special issue.

Mills

Yes, there’s a long discussion of the basic design -

Sullivan

The American Proc IRE?

Mills

Yes. And, there’s a long discussion of all the basic design factors which were considered and things like that in that paper. So we had a minimum size from the astrophysical requirements. The maximum size wasn’t so much money, but the amount of effort we were prepared to put into it. And I had originally hoped to have it a bit larger, something around about half a degree, but it became pared down a bit with time until we finished up with a beam width of about 49 arc minutes.

Sullivan

You say the frequency was chosen for maximum sensitivity and you were at 3-1/2 meters, but why not go to a lower frequency?

Mills

In that paper I mentioned earlier, I have a discussion of this – at the moment I can’t quite remember what all the various factors were. It was connected with resolution, too, of course. If you go to a lower frequency the antenna has to be larger. You get a balance between resolution and sensitivity, particular frequency, particular receivers. There was a very wide maximum but roundabout 80 MHz seemed to be about the right frequency to do it.

Sullivan

Were you thinking at all in these years that by observing at only very low frequencies that you might be biasing the nature of the objects you were finding – that there might be lots of other types of sources -

Mills

No. It was purely (?)

Sullivan

Which you would pick up at 800 MHz or something?

Mills

Yes, well it was known that the sky would undoubtedly look different. We didn’t at that time, I think, seriously consider that there would be sources stronger at a high frequency than they were at a low frequency.

Sullivan

Although Taurus was sort of puzzling.

Mills

It was more or less flattish. Everything we knew, except for HII regions, was falling off at high frequencies. The reason we chose a low frequency, of course, was simply that the receivers in those days – the high frequency receivers – were very poor.

Sullivan

Can you tell me about your six-month trip away from Australia? Where did you go?

Mills

This was spent in two places – at Caltech, which was the basic reason for me going there, to learn some basic astronomy, which, I felt, I was sadly in need of in those days. Caltech was the center, of course, and this was the obvious place to go to talk to Minkowski and Baade. I’d been corresponding with both of these. And to look at the sky survey plates.

Sullivan

Which was in progress then?

Mills

Yes.

Sullivan

When was this?

Mills

This was in ’53, about mid-53. It was shortly after the publication of our first Cross paper. From my point of view this was a very fruitful part of my life because I could stop worrying about instrumental developments and sit down and start thinking about astronomy, astrophysics, and physics generally. I spent about three or four months there and another three months or so at the Department of Terrestrial Magnetism, which I looked on more as being in the nature of paying for my visit since they were just becoming interested in Radio Astronomy. And I, in fact, did help them out on a lot of technical matters. It was also very valuable too, of course, to work in different groups.

Sullivan

This was with Tuve.

Mills

Yes, Merle Tuve.

Sullivan

And was anyone else with this?

Mills

Well, Graham Smith was overlapping me there – he spent some time there.

Sullivan

I guess you were the one that probably influenced them to build the Cross, and I think Graham Smith actually finished it up.

Mills

Well, no. Actually they had seen my paper before I came over, and had decided to go ahead and make a Cross. So that when I came over, in fact, they were already starting on this. I didn’t really contribute very much on that Cross part at all. The decision had already been made.

Sullivan

I’m interested about your visit to Pasadena. Can you tell me what Baade and Minkowski were thinking at that time? What they were teaching you, essentially?

Mills

It was all colliding galaxies in those days, of course. There were a lot of discussions about probability of collisions, and how many there should be. There were some worries because there didn’t seem to be enough collisions for the radio sources which were observed. There was very little physics in that part of it. It was purely looking for abnormalities in the galaxies which might be identified with radio sources. But how the physical connection, no one had a clue about. The mechanism wasn’t known then. I think I mentioned before I’d had some suspicions that it was connected with synchrotron emission after reading Fermi’s paper on acceleration of cosmic rays. And there were one or two papers, I think Herlofson produced one -

Sullivan

Alfvén and Herlofson and Kiepenheuer.

Mills

Yes, and Kiepenheuer too. Although somehow I missed his. I don’t know how I missed Kiepenheuer’s paper.

Sullivan

This was back in 1950. They were both in Phys. Rev. within a couple of months of each other.

Mills

Yes. Well, it was odd but I was aware of the Alfvén-Herlofson one. I had a sort of feeling this was probably the reason, but I didn’t know how enough astrophysics then.

Sullivan

Did you know about the Russian work that was going on in ’52 and in ’53?

Mills

No, I knew nothing. I didn’t strike that until later. I can’t remember where I first struck that, but as soon as I saw it, everything clicked. That was it.

Sullivan

You were a believer then?

Mills

Yeah, I was a believer. (laughter)

Sullivan

So it sounds like from what you’re saying that Baade and Minkowski were really the only two at Pasadena that were interested in radio sources. Is that fair to say?

Mills

Yes, I think so. Greenstein had a bit of an interest, too. But they were the two fundamental people. As I said, we were all searching. Then, as to just -

Sullivan

Was Hubble still alive when you were there? He died in ’53, I think.

Mills

Yes, I met Hubble actually. I had a short conversation with him and he died while I was at DTM, I think.

Sullivan

Was he going along with this or was it outside of his realm?

Mills

He was interested in an off-hand sort of way. He wasn’t professionally interested. Whereas Baade and Minkowski were really working hard on it, and devoting all their efforts at that time to these problems. We just didn’t know – it was a question then of, for instance, whether Population II objects were connected with it because of M87, an elliptical galaxy. Or whether Population I -

Sullivan

Because of all the dust and Centaurus?

Mills

Yes. This was one of the things which worried me because I had this tendency towards thinking in terms of synchrotron mechanism – just in the back of my mind – but it seemed to me M87 was the thing which really killed that. And also 5128.

Sullivan

Because it looked like such a tranquil galaxy?

Mills

Yes. 5128 also appeared to be an elliptical. So it looked as if these were elliptical galaxies and in fact, as if these might be associated with the radiation in old stars. These were a lot of the things we discussed at the time. One of the first things I did with the Cross when it was going (if I can anticipate a little bit) was to look for emission in globular clusters to see if that particular problem could be resolved.

Sullivan

What about Hanbury Brown and Hazard’s work on showing that spirals had strong continuum radiation?

Mills

Well, we knew the Galaxy did.

Sullivan

Well, M31, then they began to pick up more and more spirals.

Mills

I can’t remember the exact dates of these. This picture was fairly clear that we had normal galaxies like the Milky Way which we knew produced emission. M31 produced it. Others you’d expect to produce it. But these were thought of as normal galaxies. What we were worried about was this very abnormal emission.

Sullivan

Right. And were they beginning to be called “radio galaxies” then?

Mills

I think they probably were. I can’t remember exactly when the term arose. You’d have to look through the literature to find that.

Sullivan

Well, this probably explains the source of your short article in Observatory in ’54 entitled “Abnormal Galaxies as Radio Sources.” You talk about Fornax and you say, “5128 may not really be colliding galaxies.” I guess that was just what you’ve been saying now that you had this feeling.

Mills

Yes. This was written after I came back from my American visit, and after the Cross had started working and these were some of the first results. I was working through some of the ideas which I had picked up in Pasadena. I can’t really remember that particular paper. I know that it was written about that time, but I can’t remember what was in it.

Sullivan

Do you think it’s fair to say that the collision hypothesis was sort of grasped on because there didn’t seem to be anywhere else to get such an amount of energy – kinetic energy?

Mills

Yes, that was it, certainly. And, in fact, there was no doubt in ’53 between Baade and Minkowski that they were dealing with colliding galaxies. In fact, Minkowski paid off a bet of a bottle of Scotch to Baade as soon as they got the spectrum of Cygnus A.

Sullivan

Right, I’ve heard that. When did they become converted on that score?

Mills

I can’t remember the exact sequence, but more information was coming in all the time and I think, in fact, 1316 and 5128 did go a long way towards checking them on this. Because it was very difficult to really get a consistent colliding picture. And there was also another northern multiple source, I can’t remember which one it was now, which Minkowski originally thought of a colliding but then he found several different velocity groups in it, and this just looked very peculiar. I mean, you can have two velocities with things colliding, if you’ve got more, it suggests something expanding.

Sullivan

Was this Perseus A?

Mills

Yes, it was.

Sullivan

Of course we still don’t understand what’s going on.

Mills

That’s true (laughter). But, I think that was, perhaps, the first thing which shook Minkowski.

Sullivan

But why do you say Centaurus didn’t fit in? Because when you took spectra you didn’t find two sets of lines? I don’t remember in Centaurus whether you do or not.

Mills

The disposition of the lines, I think, didn’t really suggest two separate colliding systems. I can’t remember the details of it.

Sullivan

Ok, moving on - a short article in Observatory in ’56 on the radio source near the galactic center, which you apparently were interested in. What was the feeling about this radio source?

Mills

Well, there’d been a number of papers produced by the high frequency people – NRL had done a lot of work on the galactic center source. So obviously this was something that one should -

Sullivan

Also, I think Bolton and somebody had a paper, didn’t they, on the whole galactic center region where they mapped it?

Mills

Low resolution map, I think.

Sullivan

Right, I guess you really wouldn’t see the source.

Mills

No, it didn’t show up at all. So we looked at it with the Cross. And as soon as you look at it, you saw that you didn’t have one source flat at the center, but you had two sources straddling the high frequency source. This was obviously an intensely interesting result. It was one of the few occasions I think where I sat down and write a paper very rapidly as soon as we had the contours worked out. It was clear that this non-thermal emission was arising on either side of the thermal emission. I didn’t believe some of the distance analysis which had been made at the time because it was obvious to me that this thermal source which the high frequency people found, you couldn’t really estimate the emission from it with any degree of certainty. Because the baselines – base levels – were so uncertain there. I felt this must be at the center of the Galaxy because everything was symmetrical. I went ahead on that basis.

Sullivan

There was a paper by, I think, Bolton and McGee on Sagittarius A and I think it was about ’53 or so. It was definitely earlier than ’56.

Mills

Yes, that’s right.

Sullivan

But I can’t quite remember what it said.

Mills

I can’t remember what it said either. I remember, though, it was observations using a hole-in-the-ground type of antenna with a moveable feed, but also Piddington had done measurements at high frequency too. So the galactic center was an interesting thing. I mentioned Haddock and the NRL people because at that time they had the best observations.

Sullivan

Of the high frequencies?

Mills

Yes. In the Australian group there had been observations of it before then.

Sullivan

So it sounds to me like in the mid-50’s there was quite a bit of controversy then as to what the nature of this thing was. There were arguments that it was only 2 or 3 kiloparsecs away and other arguments like yours.

Mills

Yes. As far as I remember the short distance arguments did originate in the NRL group.

Sullivan

From HI

Mills

And from HI.

Sullivan

I guess that may have been a little bit later. But I think they had a distance even before they got the HI absorption.

Mills

Yes. Based on, I think, the ratio of the emission to the background, or something like this. I can’t remember the argument.

Sullivan

Ok, let’s clean out a couple of these others before we get into the whole MHS survey. Absorption of 3.5-meter radiation in 6357 which was an HII region. According to the abstract, said that it was the first ever seen absorption, but hadn’t Ryle and Scheuer shown that the whole galactic plane absorbed, in ’52 or ’53?

Mills

No, that was an inference rather than a direct observation. And the inference was actually wrong because there were a large number of emission regions. In fact, this was the only absorption at 80 MHz, that is the only obvious absorption.

Sullivan

Anywhere in the Southern - ?

Mills

Along the galactic plane. Eventually we found others, but widespread absorption sets in at very much lower frequencies, around 30 MHz. So that, in fact, was one of the other sources of argument between Cambridge and Sydney.

Sullivan

Well, I’m going to talk with Scheuer sometime this week, so I’ll see what he says about that.

Mills

You’d have to look at the original papers to sort that one out because I can’t remember the details.

Sullivan

And you, in fact, Mills, Little, and Sheridan in ’56, you published looking at 14 bright nebulae – I guess they were all HII regions and detected 6in. emissions and got one in absorption. And you were deriving electron temperatures of 10,000 K, so it seems like you’re really getting into astrophysics here.

Mills

Yes, we were.

Sullivan

How did that come about? You didn’t learn about HII regions at Caltech, did you?

Mills

Oh, I learned about everything at Caltech. I had also, of course, been reading astronomical literature. I think the Caltech, Pasadena visit was mainly a gelling of a lot of things and gave me more of a feeling for the astrophysics of it.

Sullivan

Ok, here’s another paper in ’56 Mills, Little and Sheridan, in which you look at two supernovae and 10 novae and only detected one supernova, namely, Kepler’s. And yet you make a suggestion the galactic background may just be the (?) integral of many supernova remnants which is a rather interesting comment. Were you thinking of these remnants as actually being the source of energetic particles perhaps?

Mills

Yes. At that time the picture was simply that the supernovae were going off and what you were seeing was the integrated result of old remnants which had expanded.

Sullivan

Well, you say the picture. I don’t think everyone accepted that. This is more of the Russian line, tying in cosmic rays and background radiation -

Mills

Yes. I think around about that time I became – my outlook was quite close to the Russian. As I said, I was very impressed with the Shklovsky work and the interpretation of the Crab Nebula and things like this. I think I was just generally thinking along similar lines.

Sullivan

Another one here on looking at bright galaxies in which you detected 10 and the Magellanic Clouds – more than the detected, but mapped them out – and you made a model with a Type I distribution, I guess that meant Population I, and a spherical halo. I guess this was under the influence of the astronomy that you’d learned also, I mean about Populations and such.

Mills

Oh, yes. I was by then quite familiar with all the latest ideas in astronomy and astrophysics. There’d been a lot of work on Populations and formation of the stars in the early days of the galaxy. I remember Martin Schwarzschild was at Pasadena at that same time and he was thinking of these things, too. And there were numerous colloquia which I attended. As I said, it was terribly fruitful. I learned all my astronomy in one year.

Sullivan

That was your graduate education in astronomy, essentially.

Mills

Exactly, yes. It put me right in the forefront of thought at that time.

Sullivan

Were you thinking of the two Populations being analogous to your Class I and Class II?

Mills

No, in this model -

Sullivan

No, that’s not right, I withdraw that question.

Mills

Obviously for the Population I emission, there were supernovae and HII regions and things like this. There was also a halo then, which came directly from the observations. In the paper you’ll see some sections through the Galaxy which did suggest this. And this had been suggested by Shklovksky, of course, and I rather looked on these observations as confirming this picture of a more or less spherical halo. But the Population I component he had attributed entirely to thermal emission in his model and I think I pointed out that this was non-thermal as well as thermal. That was the difference.

Sullivan

I see. So are you saying that in the maps of these nearby bright galaxies you could actually see a hint of what we now call a disk and a halo?

Mills

No. I could see that in the Galaxy, in the Milky Way.

Sullivan

And maybe the Magellanic Clouds?

Mills

In the Magellanic Clouds there was no evidence of a halo. When you plotted the emission, you found that the spirals, the Sb spirals in particular, seem to have more emission than some of the earlier type galaxies. There seemed to be a maximum. You had irregulars like the Magellanic Clouds with low emission – the radio to optical ratio was low. Things like the Milky Way and M31 and other normal spiral galaxies seemed to be high. When you got to the ellipticals, it dropped right off again except for the odd radio galaxies. So it appeared that, this again, I thought, was dividing it into Population I and Population II objects and it did appear that you had to have a combination of them to produce the emission in normal galaxies. I discussed this to some extent in my paper on the “Observation of Normal Galaxies.”

Sullivan

And then your observations of these other galaxies was consistent with this picture but didn’t really establish the picture.

Mills

Yes. [It appeared, in fact, that strong emission requires large mass (Population II component) and a Population I component. I think that was a suggestion I made to get radio emission in significant quantities. Tap garbled and difficult to understand]

Sullivan

In mass?

Mills

Yeah, well, that seemed to be the thing that correlated with it. I think you’ll find that discussed in one of those papers.

Sullivan

We come along to Mills and Slee ’57 which is the first installment of the MSH survey which I guess was the main thing you always had in mind to do with the Cross. Or is that true? Was it your main idea to really do a super survey?

Mills

Yes. A survey -

Sullivan

But you had all these other projects on the side?

Mills

Yes. We were interested in galactic structure and the survey in the beginning and the survey, of course, takes more time. That’s why the other ones came out first.

Sullivan

Can you tell me how this survey proceeded? Was it pretty straightforward?

Mills

Yes, each night we did one strip of the sky and moved on the next night, interspersing this with observations of individual things which came out in the earlier papers. But the basic bread-and-butter work all the time – when it wasn’t doing anything else – it was on the survey.

Sullivan

Were there any ambiguities as to how this data should be reduced – not interpreted, I mean, but reduced?

Mills

There were lots of discussions as to what would be the best way of reducing it. This was determined in the end, I think, by our facilities. But I don’t think there was any real – we didn’t consider two possible ways and decide between them. It was just a matter of gradually working out what was the best thing to do.

Sullivan

Could you just describe the procedure?

Mills

I’ll try. We first set up – calibrated everything before the observation started with noise diodes and things like this. We got an observational run through. If you look at the paper, you’ll find that the actual records consisted of a series of interleaved observations at different declinations. We then employed girls to trace these out and give us a series of scans at neighboring declinations. We had five beams at a time. Yes, here’s a typical record.

Sullivan

Figure 1.

Mills

Actually some of the other earlier papers showed it better, with sections through the Galaxy and things like this which showed larger sources and so forth. But each of these scans on here, vertical lines, represents an integration and the peak value of a scan represents a value of the integrated emission over that time. And we simply switched from one declination to the other at the five declinations and back again. So that every fifth point there represents the emission at that declination. So we had to separate these out and we did this with the aid of girls, being very cheap to employ in those days.

Sullivan

And this is what you got in Figure 2, then?

Mills

Yes, that’s right. Then we looked through these looking for bumps. This is basically what we did. Most time it was obvious, sometimes it wasn’t. The ones which were not obvious, of course, caused all the difficulties and problems.

Sullivan

How did you handle them if it was not obvious? What were your criteria?

Mills

Dealing with the obvious ones first, we’d draw a baseline under it. We’d get a deflection on different declinations and a time of maximum (time of maximum would determine the RA) and by fitting these to an exponential beam – we had a little formula for doing that – the ratio between them would give the declination. So we got our positions that way and we got the flux density by drawing a baseline underneath and measuring the deflection above that baseline. That was the basic technique. And when we had troubles, when things were apparently extended, we know now that a lot of these extended ones were background irregularities, we thought at the time they might be, but we included all these in the catalogue because again we didn’t know what we had to look for at that time. And any concentration of emission of any sort appeared to be worth cataloguing. And we were not necessarily thinking of cataloguing only point sources, but all concentrations of radio emission.

Sullivan

Once again it was sort of an operational definition – “a bump on the record.”

Mills

Yeah, that’s right. If a bump were extended, we would proceed in the same way and obtain an integrated flux density rather than just a deflection and again give a position and we’d give the ratio between the point and the integrated flux density.

Sullivan

Did you think about trying to make a little contour map if they were extended?

Mills

We did think of it and we, of course, did so for specific sources, but we didn’t in our catalogues. It was just too much labor involved. All this was manual reduction. No computers available at all.

Sullivan

No computers even up to the late ‘50’s?

Mills

We didn’t have any available to us. The one which had originally been developed at CSIRO in radio physics, which I mentioned earlier, was abandoned. It was decided that that wasn’t an appropriate field for radio physics to deal with and so that was sent away and we had none. Computers didn’t become available in Sydney probably until about ’54, but they weren’t available to us then.

Sullivan

Now, these extended sources, of course, are critical to the whole argument. Were you thinking that a goodly number of extragalactic sources might well be extended also, because you were thinking of Population II and this sort of thing?

Mills

Well, it was obvious these could be either galactic or extragalactic. We tended to think that most of the extended ones were probably extragalactic, and now this is probably wrong – they were probably galactic, I would say. And either way, of course, they must be included in any statistical analysis. Account must be taken of them because they are there.

Sullivan

Well, something must be done about them.

Mills

Yes, that’s right. One could eliminate them entirely or else – but since one didn’t know whether they were galactic or extragalactic, the reasonable thing to do is to do statistics on them and see how these statistics turned out. And see if that would indicate anything, which is what we did.

Sullivan

Were you worried at all about blending of two point sources, essentially?

Mills

Yes. Well, we knew that some of these would be blended sources and we worked out the statistics of that.

Sullivan

As to how many could be?

Mills

Yes, and we found that there were many more, so we thought there may be extra blending. You see, at that time, we knew about radio galaxies, we knew there were clusters of galaxies, so we thought there was a distinct possibility that there would be physical associations between several radio sources. So we felt that a lot of these extended things – you see, our basic resolution was ¾ of a degree, and we couldn’t resolve very much. So we thought it was highly probable that some of these extended sources did, in fact, represent blends of radio galaxies.

Sullivan

This survey began about when?

Mills

It would be ’54 that it actually began.

Sullivan

Well, you talk about Pawsey going to the Jodrell Bank Symposium and appearing to contradict the Cambridge observations. Was that the first inkling that you had that things were looking rather different in the two surveys?

Mills

Yes, that was the first we heard of the Cambridge 2C Survey – at the time of this report of Pawsey. I don’t know when that was, ‘55, yes.

Sullivan

And was this a very small section of the total survey which came out – submitted in late ’56. How did this come about? I can’t quite remember – did you see the published 2C catalogue?

Mills

No, not at that time. I just remembered something. Actually, before we heard from Pawsey, I got a letter from Fred Hoyle asking or mentioning a very great excess of faint sources which they had obtained at Cambridge, giving a slope of 3, minus 3, and asking what did we get. We had rough results then. It was obvious we couldn’t possibly have a slope of minus 3, so I wrote back saying this. And then this was formalized more in the Jodrell Bank Symposium that Pawsey attended. But, the first indication, I think, was a letter from Fred Hoyle.

Sullivan

I see. Do you think that letter could be dug up?

Mills

No, it doesn’t exist, I’m sure. Well, it might be possible. I’ve got some old files there; I’ll dig through them.

Sullivan

I’d appreciate it. But anyway, so that was your first indication that you were getting rather different results. And how were you able to make the detailed comparison? Did you have that before publication?

Mills

Yes, I think Pawsey organized this while he was over there. They sent over a section of the catalogue and we reduced – I’m not sure how this was organized, whether I had actually told Pawsey, “Well, we’ve nearly reduced this area, see if you can get something on that.” I think that’s actually what happened. Anyway, we had this result and -

Sullivan

It wasn’t published yet?

Mills

Ryle may have sent it, I can’t remember. We certainly had it before – it had been submitted at that time. All the work was completed. It was just a matter of sending a sort of preprint of the thing.

Sullivan

And what did you think when you had a few hours to look at this?

Mills

It was obvious that Cambridge were not listing radio sources as such. Because one could take an area where there were some quite strong sources in the Cambridge catalogue, look at our records, and see that there was absolutely nothing there. And so it had to be explained. Now I was already quite familiar with the effects of resolution then. So the explanation was obvious. We did a bit of simple figuring on the beam areas available and it was clear that they were listing about one source per beam area.

Sullivan

Right, which you point out in this paper. Of course, you explicitly show the 2° by 15° Cambridge beam and your 1° beam approximately. You say it was obvious, but how did the Cambridge group go wrong?

Mills

I really cannot understand that. They knew about the resolution problem, I’m quite sure, at that stage. In fact, I had discussed this with Graham Smith way back in ’52. How they could possibly have interpreted their records as being isolated discrete sources, I simply cannot understand. You ask them. (laughter)

Sullivan

Well, I have, but I’m trying to get opinions all around. I don’t know, maybe it was something to do with a sort of hunger for a great number of sources to do cosmology or something.

Mills

Yes, that could be.

Sullivan

There’s more in science than science sometimes.

Mills

Yes, if one has some preconceived idea, then an observation which apparently confirms it will often be accepted without very great criticism.

Sullivan

I don’t think that’s fair to say that there was a preconceived idea because it’s my understanding that when Cygnus came along… Well, I should ask you before I say what I was going to say. When you started off on this survey, were you interested in cosmology? Did you see this as something useful that could be done with the survey?

Mills

Not when I started on the first one. That’s the ’51 survey.

Sullivan

Not that one. Right. But, I meant with the Cross.

Mills

By then I knew that one had a chance of doing cosmology. I certainly knew that when we started with the Cross survey.

Sullivan

But what was the reasoning of it?

Mills

Well, you were dealing with distant sources and they were at cosmological distances. The identifications showed that they were very, very faint things. Strong sources were very faint optical objects. Obviously it was relevant to cosmology. We all knew that.

Sullivan

The Cygnus identification really is the key to it, being the weakest optically.

Mills

But on the other hand, we knew that 5128 and M87 probably wouldn’t have very much cosmological interest. And it was just this appreciation that there was undoubtedly a very wide luminosity function.

Sullivan

So when you started the MSH survey then, you were thinking that it would have a cosmological use?

Mills

That it could. I didn’t have any ideas that it would turn out to have such an enormous cosmological impact from the Cambridge viewpoint.

Sullivan

Did you actually sit down beforehand in terms of the design of the survey at all and say, “Now, in order to test this cosmology or that, I need so many sources per - ”?

Mills

No, not at all. The design of the instrument and the planning of the observations was based on what we had observed before in the radio sky entirely.

Sullivan

I guess you felt compelled to come out with this small part of the survey.

Mills

Yes, we did. We had some correspondence about this and Ryle was quite unshakeable and I know Pawsey had argued with him about this. His opinion was that Ryle, again, was quite unshakeable. (laughter) So we felt the only thing to do was to do a thorough analysis, as far as we could, of comparing the two catalogues, as being the only way of sort of bringing things to a head and establishing -

Sullivan

Any of that correspondence would be extremely interesting, if that could be located.

Mills

I’ll see if I can dig anything out. But as far as I remember it, my efforts were met with no reply. I remember sending one letter to Ryle, for instance, saying, “Well now, here are our records of this area where you have these sources shown. There are no sources there. Could it be resolution effects?”, or something like this. And I never got an answer. We were a bit fed up with the Cambridge attitude at this time, I might say.

Sullivan

Well, it comes through in this paper.

Mills

They just ignored us. (laughter) So we went ahead and did what we felt had to be done.

Sullivan

Well, related to an earlier question I had, do you think it possible that if you’d been working at Oxford that this could have gotten straightened out?

Mills

I think it probably could. Although, I think Jodrell Bank had difficulties straightening things out with Cambridge, too. (laughter)

Sullivan

Well, they never conflicted so much on direct observational things.

Mills

No. I’m sure if we’d been able to take our records along and put them down on the table and look at their records -

Sullivan

But now someone had told me, I think David Edge told me, that you actually came up and did this in connection with the 1958 Symposium. This is getting a bit later.

Mills

Oh yes, that was later. We did that then. At this earlier stage, our results were, as far as we could determine, just completely dismissed by the Cambridge group. So we had no alternative but to come out with this sort of paper.

Sullivan

What was the result when you finally did get up and were able to lay your records out?

Mills

I think by then the Cambridge people were fighting a rearguard action on the 2C survey, and were starting off on the 3C. I think they appreciated the detailed criticisms. I’m sure they had all abandoned it by then.

Sullivan

So this was mainly to look at 3C.

Mills

Yes, that’s right.

Sullivan

Well, as you’ve already hinted at, this paper doesn’t mince words – “It is found that the survey is almost completely discordant,” a couple of other good quotes in here. But the paper largely seems, rather than talking about your survey, I mean you do analysis of it and so forth, but rather it’s much more emphasized towards showing that 2C is wrong.

Mills

Yes. This was necessary at the time, I think. Because there was enormous publicity attendant on the 2C.

End Tape 62B

Click start to listen to the audio for Tape 63A of the 1976 interview.

Begin Tape 63A

Sullivan

Continuing with Mills on 26 August 1976. You were mentioning that there was quite a bit of publicity about the 2C survey.

Mills

The cosmologists appeared to jump on it, of course. The general picture that came back from the ’55 Jodrell Bank Symposium, as relayed by Pawsey – and I think Bolton was there too, if I remember correctly, I’m not sure about that, but I know Pawsey was there – was that it was creating enormous interest and if it were taken seriously, of course, the implications were really very strong. We knew that it was quite strong. I had perfect confident in this, although one didn’t have any great confidence in some of these weak sources that we had. It was quite obvious that the Cambridge 2C survey was strong at a very much higher level than ours could possibly be. So it obviously had to be straightened out. We couldn’t get through to the Cambridge people and this came strongly, as reported by Pawsey. And of course I discussed this with him again as to what was the best thing to do, and we decided on this critical paper.

Sullivan

Did Pawsey bring along strip charts and try to - ?

Mills

I can’t remember whether he brought any back. He didn’t have any of mine.

Sullivan

That’s what I meant.

Mills

No.

Sullivan

But no you also mentioned that Ryle had a popular article in Scientific American about this thing and you felt compelled to write a Letter to the Editor.

Mills

Well again, I think it was Bart Bok who suggested I do that. I’m actually not too happy about getting into public controversy, but this was one thing which was obviously terribly important. I felt, and other people felt, it should be straightened out. Hence, both the reply to Ryle in my letter to Scientific American and bringing out this paper. Though this paper was well advanced, of course, at the time of that letter and exchange in Scientific American.

Sullivan

Right. This was submitted in the same year as the other one was published. Now did this survey get a lot of publicity in Australia or elsewhere?

Mills

Oh, yes. People were using it quite a lot, I think, at that time. But it’s known publicity came as contradicting the 2C, I think.

Sullivan

Right, but that’s what I mean. Did it get fair press?

Mills

Yes.

Sullivan

Ok, let me just got through the paper and ask you a couple of specific questions. You say, “Reliability of better than 90 percent of greater than 20 Janskys (as we would now say) – these should all be included and completely reliable.” Your catalogue actually has a greater number of sources between 7 and 20 Janskys.

Mills

Yes, that’s right. Again, the purpose of the catalogue was to produce a catalogue of all concentrations of radio emission and this was the thing which we were aiming at. We were not aiming at making sure we were cataloguing only radio sources as such, point sources.

Sullivan

So you were just giving the reader the whole bit of data and then just saying -

Mills

Yes, “I would choose to cut at about here.”

Sullivan

You do a lot of analysis here on the effects of confusion in extended sources. And you end up saying that some 90-odd percent of yours are reliable and 98 percent of the Cambridge ones are not reliable, which, not quibbling about a few percent, basically turned out to be right.

Mills

Yes.

Sullivan

And here’s the great difference in log-n/log-s. Although I think that even your 20 Janskys in retrospect turns out to be a bit optimistic.

Mills

As far as the point sources were concerned, it was ok. But there were a lot of extended sources which, in fact, turned out to be background irregularities and some, I think, though, I don’t know how many, were, in fact, background irregularities.

Sullivan

You talk here about clustering of sources and you think that you have a .02 result here, a 2 percent chance of it being random. Was that just a statistical fluke?

Mills

I think so.

Sullivan

Did you try to do this later with the entire survey and it didn’t pan out?

Mills

Yes, I didn’t get any significant results. It was just a fluke.

Sullivan

And here you talk about the problem of the background variability, and you’re trying to estimate how much noise this would add.

Mills

No, the basic idea was in effect a crude “p of d” analysis. All I was doing was measuring the rms background fluctuation and interpreting it in terms of the integrated effect of many sources. And this led to this sort of cosmological model.

Sullivan

And you come up with a density that turns out to be about a thousand times that of Ryle and Scheuer. I figured that out.

Mills

I think I suggested one possible cosmology which would fit, but I realized all this must be terribly preliminary and it was just a matter of seeing if it was reasonable or not, which it was.

Sullivan

Yes, here’s: “We therefore conclude that the discrepancies in the main reflect errors in the Cambridge catalogue and accordingly deductions of cosmological intent derived from its analysis are without foundation.”

Mills

Yeah.

Sullivan

But, now here’s something philosophical – “An analysis of our results showed that there is no clear evidence for any effect of cosmological importance in the source counts.” But that’s not true because -

Mills

No, it is true.

Sullivan

If you say that they’re extragalactic and you have -

Mills

Of cosmological importance. There were cosmological effects. But I don’t think – we couldn’t demonstrate anything of cosmological importance in our counts.

Sullivan

One could argue that this established that these were a Euclidean Universe.

Mills

No, we merely say that we didn’t go far enough to distinguish between different universes. Every universe you start off counting looks Euclidean. And it’s only when you go to weak sources that you get cosmological effects coming in. This was my thought.

Sullivan

I see. So through your analysis you had determined that you really didn’t think that you were going deep enough, as we would say?

Mills

Yeah. Or, there may have been a population of more local objects, too.

Sullivan

Still extragalactic but - ?

Mills

Possibly. Anyway, local shall we say. One couldn’t distinguish in those days. You could say everything was distant extragalactic objects, but then we knew that some were not. We knew that M87 and 5128 were quite close extragalactic objects. We had no clues about the luminosity function. There were not sufficient identifications to give a hint, except for Cygnus A which suggested there were some very strong ones. These may have been a very small proportion. And if things were all like M87 and 5128, one wouldn’t have gotten any cosmological information. We got nothing statistically different from a Euclidean slope and the fluctuations in the background were such as could be explained by a fairly local population of extragalactic sources. So that we didn’t have anything really significant.

Sullivan

Did I hear you right though, when you said that there was still, in the back of your mind, some thought that they just might all be local within our Galaxy?

Mills

Not all. There could be a dilution.

Sullivan

Or a goodly fraction?

Mills

Yeah, I still kept that as a possibility at that time. I didn’t have any strong feelings about the necessity for them being cosmological or local at any time, I don’t think. Not until we started doing lots of identifications and it became clear you could account for practically everything by extragalactic sources.

Sullivan

Another technical question. Were you aware at this time that the steady state universe, in fact, would not predict 1.5 for the slope?

Mills

If you take them sufficiently local it does!

Sullivan

I mean if you put in the redshift effect, for instance?

Mills

If you’re looking at distant sources, again, no cosmology with redshift would predict 1.5 and I knew that.

Sullivan

Ok.

Mills

But it was a question of where this turnover occurred. That was the crucial thing.

Sullivan

I guess the thing that I’m learning from you that I didn’t realize is that you were really thinking that they were probably extragalactic, but that you weren’t going deep enough to do cosmology.

Mills

Yes, that was my thought.

Sullivan

I hadn’t quite seen that in the paper before. And, therefore, you only say a couple things about cosmology. You don’t try to put forth a whole cosmology of your own.

Mills

Yes.

Sullivan

You brought up “p of d.” Now, the Cambridge people, in general, will say after two or three years they began to have to admit 2C was pretty bad. But they hung on to “p of d.” It seemed to be right and it seems to me, and I haven’t gone into every bit of analysis, that indeed history has shown that it was basically right. Would you agree with that, first of all?

Mills

Well, the measurements were obviously right, but they were open to several interpretations in those days.

Sullivan

But I mean that the whole concept of a “p of d” analysis was a valid procedure.

Mills

Oh, yes. I think Minkowski put that very well when he said that “The ‘p of d’ carried information, but very low-grade information.” And that, in fact, was our feeling.

Sullivan

Was that at the Paris Symposium?

Mills

No, that was a private comment to me. He visited Sydney actually about this time we were finishing off this paper, and some of my conclusions there probably reflect my discussions with Minkowski about these things, particularly on the confusion problem. I know we had a number of discussions about that. Anyway, that was a bit of diversion from your actual question.

Sullivan

It seems to me that the outside world looked at “p of d” as sort of a rearguard action, trying to save what had been years of work, etc. Although it also seems to me that, because it was so complex, the outside world didn’t really give it the chance it deserved, to find out that it could establish some things.

Mills

I think we had a correct view of the importance of it. That it carried a lot of information, but I think Minkowski was right. It was low-grade information because you had to put a lot of assumptions into it. If you assumed that sources were discrete point sources randomly distributed through the universe, it carried an enormous amount of information. But at that time there was just no reason for assuming this.

Sullivan

Ah, yes, you were still trying to establish that?

Mills

Yes. Only Cambridge knew this, no one else did. (laughter) It was an assumption that went into “p of d” and as far as that assumption went, its conclusions were valid. And I quite accepted that. There may be a bit of discussion in the Paris Symposium about that on my part because if the sources were not randomly distributed, if there were groupings -

Sullivan

And you had some indication of clustering.

Mills

Yes, I did. Then the “p of d”, all it can do is set an upper limit on the population, because clustering increases the probability of deflection. I was aware of this, so that this excess, which was also indicated by the “p of d”, could also simply indicate clustering. Again, I discussed this with several people and I think Minkowski and I were in fairly good agreement on this. I had spent a lot of time with him on it. I didn’t at all reject the importance of “p of d” analysis then, in fact, to some extent, still feel that one has to be very cautious interpreting what comes out. Then, and I think I would still say, “p of d” sets an upper limit on the distribution of point sources.

Sullivan

Above a given flux level?

Mills

Yes. But I don’t think it can really set a lower limit because of the possibility of new populations coming in, and the possibility of clustering on some scale, which one hasn’t been able to investigate. It’s all an extrapolation from what one knows. And you should accept it with a grain of salt. That was my very strong view in those days and, as I said, to some extent I still believe this.

Sullivan

You’ve mentioned the Paris Symposium. I’ve heard from many people that that was a rather interesting meeting, to say the least.

Mills

Yes, it was (laughter).

Sullivan

One thing that interests me is that it was really the last time that virtually all the radio astronomers in the world could get together at one meeting. So you had a complete overview of the field. What struck you, as one of the main participants, as to the nature of that meeting?

Mills

Of course, I was particularly interested in the galactic and extragalactic work – not the solar, so we won’t talk about that. I think it settled a sort of way of outlook. I found that most people were now in agreement on what should be done and the problems of observation. Although Cambridge was fighting rear-guard action, it was recognized a rear-guard action. Their new survey was obviously going to be a lot better. And really the ideas of synchrotron emission had become clear. I felt it was really, sort of represented the peak of achievement up to that time. All these ideas were now making sense. A lot of the earlier uncertainties had been completely resolved. There was this little bit of argument with Cambridge but it wasn’t, in my view, a serious argument at that time. The really serious one was on the reliability of 2C, and that they had quietly dropped by then. And I was quite prepared to go along with arguing mildly about some of the others which were not so clear. So I felt it was a very effective and very worthwhile meeting which sort of summarized the present state very well.

Sullivan

But what about optical identifications though? That was an unsettled question still.

Mills

That was an unsettled question, yes. Dewhirst had started doing some at that time, Bok reported some, but there was nothing really systematic. In fact, immediately following that meeting, I went to Pasadena again and spent two months doing identification work on a combination of our survey plus the 3C results which were then available. I used 3C for right ascensions and my positions for declinations. And that is written up in the later paper you mentioned, about ’61, on the identification of radio sources. It took rather a long time to write up. I sometimes was rather tardy about writing up results when I knew what the answers were.

Sullivan

I don’t think I have that one because my bibliography ends in ’60. So this is one by you alone?

Mills

Yes. You did find it somewhere on your list yesterday, I think. The identification of radio sources. It was in the Australian Journal of Physics.

Sullivan

I’ll check it out.

Mills

And I think it was quite important because it was really the first time that a proper luminosity function was produced. It wasn’t a terribly accurate one then, of course. No, you had some data giving the number of identifications, the identifications which clusters and things like this, because I looked for all of them, the Abell cluster identifications as well as the radio sources. It was a systematic attempt and Bolton and Minkowski had also been working. They’d started a bit later, but they published about the same time giving essentially similar results. They all came out about 1960, or early ’61, I’ve forgotten when exactly.

Sullivan

But wasn’t it discouraging to people like Minkowski that such a low percentage of these things were being identified?

Mills

I think it was mainly challenging rather than discouraging. (laughter)

Sullivan

Dewhirst has told me that he went over and spent 6-9 months looking at all these 3C fields and coming up with such a small percentage of -

Mills

Of course, one of the troubles with 3C, even then, was that there were large numbers of lobe shifts and a lot of the positions were quite unrelated to the actual.

Sullivan

Well he looked ± one lobe shift.

Mills

Yes. I think we may have overlapped at the time. He may have been there a little before me. I think we overlapped for a time. But, certainly I was quite happy with the results I got out of it and I think, I as I said, both Bolton and Minkowski continued on, later one, with even more and better positions and got basically similar results. And a lot of the later identifications were summarized in that first paper. There was a horrible mistake, incidentally, in the Appendix, but never mind, you’re not interested in things like that.

Sullivan

What?

Mills

A mistake, a mathematical slip.

Sullivan

In the Appendix of which?

Mills

Calculating source counts in various cosmologies – in my identification paper. It didn’t alter any of the basic conclusions. It was just a source of embarrassment later on.

Sullivan

Back to the Paris Symposium. You also talked about the corona of our Galaxy, thermal – non-thermal emission and so forth, Class II radio sources being non-thermal. I would gather that you would say this was beginning to become settled in your mind.

Mills

Yes. I think by then I had an overall picture of radio galaxies being the Type II sources that I had, and supernova remnants and the HII regions being the Type I basically. The galactic corona I had sort of grew out of Shklovsky’s suggestion, but the model which I produced was quite different and quite different from the one which Baldwin produced earlier. It was much smaller and flatter.

Sullivan

How did that come about?

Mills

I was using southern hemisphere data around the plane rather than northern hemisphere data around the pole. It was just different data. But they were not all that different. Everyone seemed to agree at that time there was a lot of emission at high distances above the plane at high-z distances. And this was questioned later on and it’s clear my separation probably gave too much to the high-z radiation.

Sullivan

But, like you say, there was general agreement that there was a halo in some sense, and you were arguing about detailed shape -

Mills

Yes.

Sullivan

About the whole controversy over 2C, would you agree with this, that there was a tremendous mixture of arguments going on? There was the validity of ‘p of d’ as a check which was independent of confusion effects. There was the level of confusion in the two surveys. Percentage of extended sources and how this would influence N(S). Different Fourier frequencies, for example, and how this influenced N(S). Then steady state vs. evolving cosmologies got dragged in, and what should be counted as a source. Often the arguments got cross circuited – short circuited.

Mills

Yes, it was a very confusing period, really. I think we all had our ideas one way or another, and it’s pretty hard to work without some working hypothesis. But the arguments were only heated when it was a question whether 2C was basically right or wrong. Now that was one which was settled. But it was a very heated argument about that at one time. And there was also a heated argument on a question of whether the steady state could be shot down by the radio data. And there I tended to feel for a very long time that it couldn’t be. That there wasn’t sufficient precision in the radio, or the populations involved in the radio information were not sufficiently defined to shoot down the steady state. Well, you know what’s happened since then. But at that time we were really very ignorant about what class of source we were dealing with.

Sullivan

So you would say that those were the two most heated arguments?

Mills

They were the heated ones, yes. Because of personalities, basically.

Sullivan

You also made a survey with the Cross of the galactic plane which was Hill, Slee, and Mills in ’58. And what did this show?

Mills

The interesting things that came out of that, of course, were the steps in the distribution suggesting tangential arms.

Sullivan

That’s right, I forgot that was the paper that… yes. That was in the Paris Symposium also?

Mills

Yes, that was mentioned there. I produced a spiral arm model there which actually hung around for quite a long time afterwards. Because, in fact, the radio steps did correspond to regions where one was looking along the major spiral arms.

Sullivan

Were you being influenced by all the HI work going on?

Mills

No, not at all then. That particular interpretation was purely based on looking at my results and seeing, “We’ve got steps there. They must be spiral arms.”

Sullivan

I see. It wasn’t talking with Frank Kerr at tea or something like that?

Mills

No, not that one. So I sat down and read up about spiral arms and saw that the one possible interpretation was a logarithmic spiral, a good thing to fit, we tried to fit it, and it did.

Sullivan

But did you know about the HI work?

Mills

Oh, yes. I knew, of course, that spiral arms -

Sullivan

And this whole method of getting distances - ?

Mills

In those days the HI spiral arm map was just an unholy mess. It was not circular -

Sullivan

It still is.

Mills

It still is. Well, if you’ve looked at some of the old maps, you would realize that in those days it was even worse. I think many people felt that the interpretation of the distance was very difficult and this occurred to me as being a possible way of getting a model without that. If the spiral arms were basically logarithmic spirals, then all you had to do was to fit the tangential directions and you had it.

Sullivan

But, of course, that was a big if.

Mills

Yes. Of course it was.

Sullivan

At least in one case of our own Galaxy. You mentioned that the Paris Symposium was a rather nice synthesis of many things that had been going on over the previous years. At that time, what did you see as the main thrust that needed to be done for galactic and extragalactic radiation? What direction needed to be traveled?

Mills

For the extragalactic work, clearly the identification work was the essential thing. I think everyone believed that at the time. It was, in fact, why I went back and did it, or did some of it. That, I felt, was the most important on the extragalactic side. Obviously higher resolution was a real basic necessity, too, and I was thinking about a larger instrument which would help to overcome – to give more precise positions and better resolution.

Sullivan

This is what eventually became the Molongolo Cross?

Mills

Yes, that’s right. There was a vague suggestion by then that high frequency work was going to become important and again we were thinking of what is now the Parkes dish, a big steerable high frequency dish. These seemed to be important, mainly for the extragalactic work. But also for the galactic work, it was clear that high frequency was extremely important and again one wanted high resolution. These were both reflected by the Parkes reflector and our Molongolo Cross. These represented continuing along our present path and doing what we felt had to be done. And on the galactic side, I wasn’t so terribly interested in the Galaxy then. This became a thing that interested me more later on.

Sullivan

About this time, I believe, that something had to give in terms of big proposals for CSIRO, is that right?

Mills

Yes.

Sullivan

Can you tell me about that?

Mills

We had the three proposals around ’59-’60. Paul Wild’s ring-type instrument, for a big Cross, and for the large steerable reflector. That had been on the books ever since ’52-’53 and the chief protagonist for a large steerable reflector was Taffy Bowen, our Director. It was obvious this was necessary. I didn’t argue about that. This was number one priority and we simply had to have that. But when funds were solved for that – our available funds – very little was left for anything else. Again, solar astronomy obviously needed more instrumentation so there was a very strong case for Paul Wild’s instrument. Again, I couldn’t really argue with it. That really should be second priority.

Sullivan

You’re saying you didn’t try to argue?

Mills

I think, perhaps, there were some basic arguments whether it was worth pursuing solar astronomy at all. Give that it was worth pursuing, obviously you had to make that decision and, in fact, he made it, that solar astronomy had to continue. I would have been quite happy to see it not, but of course, that was just my own personal view. So, therefore, there was very little chance with our money available, and our facilities, to also continue with the Cross. Since I felt it was most important to go to a very high resolution and sensitivity, and my analysis then had shown that the steerable dish wouldn’t do this – at that time we had no thought that it would go much beyond 10cm and receivers in those days were pretty poor at high frequencies. Although the maser had come along and this looked as if it would make a big boost for high frequency work. It still looked as if that really wasn’t what was necessary for surveying the radio sources and precise position measurements. I felt that the Cross-type instrument was the best for that particular type of operation. I felt we had to do it, so I just looked around to where I could find some support. And I did find it in Sydney University.

Sullivan

And so when did you move there?

Mills

Towards the end of 1960.

Sullivan

Was this after all the MSH survey?

Mills

It had all been published then, yes.

Sullivan

Well, it hadn’t been published yet.

Mills

I think the last paper came out in 1960. It had all be done. It may have been ’61.

Sullivan

Right, and then you began working on the design and the funding, and so forth, for what became the Molongolo Cross?

Mills

Yes, that’s right.

Sullivan

Just very briefly, when was that built?

Mills

It started in ’62 when we got our first infusion of funds from America. And it was fully completed in ’67. The east-west arm was operating some years before that, but the north-south arm was a very difficult engineering problem – a one-mile-long phased-feed system – and that was ’67 before that was actually fully operational.

Sullivan

You bring up a point which I’d like to ask about. Namely, I didn’t realize some American money had gone into the Molongolo Cross. Was this Ford Foundation?

Mills

No, NSF.

Sullivan

NSF?! Those were the halcyon days when they were giving money to the Australians. (laughter)

Mills

In total, they gave about a million dollars.

Sullivan

But also, it’s my understanding that American money basically made the Parkes dish possible. This was the Ford Foundation.

Mills

Yes.

Sullivan

And Culgoora, was this Ford Foundation also?

Mills

No, it would be Carnegie Institute, I think, gave the Parkes money and the Ford Foundation the Culgoora instrument, if I remember correctly.

Sullivan

But why is there this recalcitrance apparently on the part of the Australian government to come forth with the goodies when obviously if you had to name, I think, one science in which Australia has excelled, which puts it on the scientific map – this is it?

Mills

Well, a prophet in his own country has little honor, I think. (laughter) And this applied exactly in this field. If some overseas people were prepared to acknowledge Australian science by putting some money into it, they were prepared to put some, too. Although, they didn’t put any into mine. This was the attitude – rather funny, but here we are. I think basically the politicians couldn’t believe that anything worthwhile could come out of Australian science. (laughter)

Sullivan

So, you are saying this is not confined to radio astronomy?

Mills

It tends to operate over all science. Although they had provided a lot of money before that for optical astronomy – in building the 74’’ telescope. They had provided funds for a high energy accelerator, Oliphant’s enormous homopolar generator which consumed a large amount of money. It could be they thought there had to be some limits in the amount. Anyway, we couldn’t interest the government in providing direct support without some stimulation from outside.

Sullivan

Well, early ‘60s is about where I’m ending things. I thank you very much for your time. That ends the interview with Mills on 26 August 1976.

End Tape 63A

Part 1

Modified on Tuesday, 06-Sep-2016 16:23:13 EDT by Ellen Bouton, Archivist (Questions or feedback)