Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Jon Shanklin by Steve Norton on 1999 October 7, Niels Bohr Library & Archives, American Institute of Physics, College Park, MD USA, www.aip.org/history-programs/niels-bohr-library/oral-histories/32994
For multiple citations, "AIP" is the preferred abbreviation for the location.
In this interview, Jon Shanklin discusses his involvement in discovering the Antarctic ozone hole. Topics discussed include: British Antarctic Survey; ozone layer observations; Dobson spectrophotometers; Halley Research Station / Base Z; Jon Farman; Brian Gardiner; Shigeru Chubachi.
Today is October 7th, 1999, just after 11:00 a.m. My name is Steve Norton and I’ll be interviewing Jonathan Shanklin regarding his involvement in discovering the Antarctic ozone hole. Just to review what we already talked about, the interview will be taped, transcribed, and once it’s transcribed you’ll have a chance to make additions, corrections, deletions, whatever before it’s archived at the Neils Bohr Library at the American Center for Physics. Okay?
Let’s begin by just getting some background — your educational background and how you came to work at the British Antarctic Survey.
In the U.K. I went to a school in a place called Chester. My parents were both geologists; I sort of followed the sciences line. I passed all my A levels, which are the U.K. exams before you go to university. Then I stayed on for an extra year to do S Levels, which are really only required if you are going to Cambridge universities. And I did reasonably well in those to get accepted at Walden College Cambridge of Natural Sciences. Natural Sciences are sort of an umbrella of all the sciences, really.
What year did you go there?
I went to Cambridge in October of 1973. So that’s twenty-six years ago. And did math, physics, chemistry, and geology in the first year, and then physics and geology in the second year, just physics in the third year, although I specialized in the geophysics side of things. Then I got quite an interest in astronomy, but the academic results weren’t good enough to do a Ph.D. in astronomy, so I went into teacher training with the view to becoming a teacher. I did what’s called the first graduate certificate in education, and after completing that I was looking around for jobs both in teaching, but also was looking at some general physics related jobs, and just happened to see a job advertised for the British Antarctic Survey in Cambridge that quite appealed to me because it involved computing physics, meteorology, and those were all the interests that I had.
And this around 197...?
This would be ’77. I applied for the job, was interviewed, and I went and had a phone call saying did I want it because the first person had turned it down. I said, “Yes, I want it. I’ll take it.”
What was the job for?
This was basically almost a dog’s body type job where I was looking at radiation charts which were being digitized, and it was doing all the corrections and working out what had gone wrong all the way down the line. Looking at meteorologic observations coming back from the Antarctic. Again, checking around to see which had been coded correctly, whether it was actually seeing what they said they saw and so on. The third component of it was the ozone observations, which at that time they just didn’t have the staff to fully work them up, so we were in the process of getting them on to computers.
Digitizing and processing?
This is really just manually typing in there as observations, and then I went along and processed them all on the computer. And we sort of work from the present backwards. We know what it was in the 1960s because it had all been worked up in Joe Farman’s book, which I guess you’ve seen a copy of.
Yeah, that was the one that was with Hamilton?
Yes, right. So it was then just a question of initiative as far as getting the radiation data all sorted out. That was fairly tedious is the only way to describe it. But if you just kept all that in, it was possible to work through it all. So we completed the three bases up until 1982 I think was the last year that I got all that sorted out. The whole thing was an ongoing process because the observations were coming in every day, so we needed to do that.
What sorts of things were you doing when you were processing that data?
Which particular? The radiation or the ozone?
The ozone data, the first task was checking has it been typed in correctly. So we had typists who were keying in the data.
So you check it against the forms that actually came back or?
No, what we did is just check that all the observations were within range of this volarad [?] radiation numbers, that the format was correct with the sequences and numbers all made sense, and that there weren’t any obvious problems. Then the next stage is to work out the ozone amounts from them so they make sense, and then work out if they don’t make sense why they don’t make sense. And there’s a whole variety of reasons from the immediately obvious ones that the typists haven’t read the writing correcting so ones and sevens are quite easy to confuse. Things like that. So just checking that the typing has been okay. But then what’s been written down may be wrong as well, and that is a particular problem for a good part of our record where the station keeps what’s called kitchen time — the time shown on the local clocks. But the observations are done in universal time, and very often they would write down kitchen time and not the universal time.
You have to go back and look up.
So you have to work out which ones they got wrong that way. So that was a deductive process. Other standard things that they did wrong, the observations get numerically coded as to what type of observation it is, and they might write down the wrong code. So a lot of the typists typed it in correctly; the actual observation time is not what they did. So you have to work it out from the actual number.
It’s like a direct sun observation.
So a direct sun observation they might have coded it as an [???] observation. And initially the two may actually give ozone amounts that look sensible, but it’s only later when you do the more detailed processing and things like that sometimes fall out and we think, “But his Langley plot was this point that’s way off the plot. I don’t know what we could have done. Oh yeah, we got most of the observation times all the way around.”
So by looking at the Langley plots you could tell what someone had coded?
The Langley plots certainly tell you if it’s coded wrong. Some of the Umcare [?] plots may plot [???] observations to get the sort of code. Again, if you’ve got a point that’s way out it may be that they’ve actually done a direct sun observation and coded it as an [???].
So if you’re doing an Umcare and you’ve got the profile you can have values —
This isn’t so much getting the profile; it’s just getting the sequence of observations because the actual dial reading should be a reasonably smooth curve. So it’s not looking at where the ozone is in the atmosphere; it’s just looking at pure observations and do the observations make consistent sense. Like if you’re just looking from observation A to B to C. If A and C are consistent where B is sort of way off and you think, “Well, if it was down low it would fit.” You can sort that out. Another way is they sometimes write down a number like 1100 when they mean 1010. After a while you get to pick out these common mistakes and can then go back and say, “Well, clearly they’ve done that.” You’ve seen them on the Dobson’s in the science museum?
No, not yet. I’ve seen them out at Boulder, though.
Right. All ours still will use human operators. We don’t have to go out and read the dial. There seems to be a tendency to get a left right confusion. So, for instance, where number ought to be where it is 980 they’ll sometimes write it down as 1020 and visa verse. So they can see that it’s about 10 and they just go the wrong way. And again those you can identify. And then there’s some that you just can’t work out what in the heck it was that they did at all, and they just have to be deleted from the data.
So measurements like that when you actually calculate the ozone values get just really weird numbers?
Yeah, and they get bent because you have to come up with something that’s so unlikely so that two or three things they might have done wrong in order to get out a sense blows them out. I think one thing that they might have done wrong, fine. Two things that they might have done wrong, if you’re absolutely certain that is what they’ve done wrong, but three things is just too — obviously it would seem that it was some unexplained error they might have brought one of the levers in the wrong position, and like that you can’t recover the ozone amount.
When you’ve gone to calculate the ozone values, have you gotten any like really ridiculously high or low?
Oh, yeah. I mean sometimes you get such a virtually zero ozone values or sort of six hundred Dobson units.
And it’s supposed to be about three hundred or so?
Well, it just doesn’t fit in. Also, they are usually doing a couple of observations at the same time, and they’ll certainly find observations a day are standard, and you need to make certain the observations are all reasonably self-consistent. Now initially, we were just doing a very rough check on all the data because we’ve got about ten years’ worth. But since then I’ve been, over the years, just going back and doing more and more detailed checks and bringing all the different types of observations into closer agreement. It’s things like that which really take a long time, because when we’re doing the instrument calibrations you can’t say that at this point the instrument calibration is so much. You need to wait until you’ve gotten another few years’ worth of data and then do best fits to say, “Right at that point the calibration was so and so.” So I would say that five years ago’s data, we’re getting pretty close to what it will finally be, but it is still room for a little bit of flexibility. The other thing is that computing powers continually increase. When we first started entering the data onto the computers they were fairly limited in ability. It would take some places half an hour just to compile a program that you wanted to use to analyze the data. And so that meant that you didn’t make many changes to it, and the sort of statistics and so on that you could do were pretty limited. Now on PC you can run the same thing in a few seconds and see exactly what’s happening to all the data, and the statistical packages are better. The long-term goal is to do further improvements before we finally lodge the data with the World Data Center. But in the meantime it’s all available in the real time for people to see when they want.
Now, you spent two years down in the Antarctic, too?
No, I probably have in total spent two years down there in Antarctic if not more than that, but I’ve never actually spent a winter down there. It’s just summer visits. The first visit was in ’81, ’82. And that turned into a somewhat longer visit than expected, thanks to Dobson’s Law, which started while I was down. So I had to sail all the way back. But that first visit down, the primary task was to take down the new Dobson data.
It was 123?
Yes. And compare with the existing one down there, which was number 31.
Well, can we take a step back here for a second?
Why was 123 ordered? Was it simply because the other ones had been out in the field so long and you didn’t have a spare? I mean it wasn’t anything like you were getting the low values.
What we wanted to do at that time was really we got the three stations: Howie, Faraday, and Great Lichen, and we had an instrument at each of them. And having an instrument at each of them meant we had to stay there because we want continuity of observations. So buying a fourth instrument that we could start a rolling program of putting a new one in at one station, bringing the one out, getting that overhauled, send it down to the next one, bring that one out. The idea was we would then have a rolling program overall so that we could be reasonably confident that the instruments were all in tip-top condition. I think probably, looking back on it, that would have worked quite happily, but I think my view now is that we really want to leave an instrument in place for at least ten years in order to get good calibration in situ, rather than having a more frequent program in exchange with the U.K. I think if we’d followed the original idea the instruments wouldn’t have been down in the Antarctic long enough to get a good in situ calibration.
Will you give me a feel for this? I don’t have a good understanding of why having them down there for ten years as opposed to five years?
Basically because we do our calibrations using the Langley plot technique. So our instruments are all, if you like, divorced from the world standard. The main reason, I think, for that is that we consider that we’re isolated. We want continuous observations. If we took an instrument out to compare against the world standard, we’d probably have no Antarctic observations for two years or something like that because that’s the length of time that it takes to exchange the instrument. Also, during the transport of the instrument because it’s going to shift and there’s going to be lots of vibration, and there’s no guarantee that the instrument that you had in the Antarctic, the instrument that you compare against the instrument you take back to Antarctic, is actually the same. I would say that’s a failing with the present system. But yeah, it’s nice to have all the instruments side by side and see if they are really the same thing. But can you guarantee that what we’re comparing is what we actually measured after the shipping? So ours, because we’re doing it with reverse principles, we’re pretty much convinced that even though we’re floating from the world standard, we will be matching that standard. Our feeling is certainly that was true when we took the, first of all, 123. We compared against med office’s regional standards.
Did you do that work?
Yes. And then my instrument, number 123, went down to Antarctica and we took 31 back and we compared that against the Meta standard. But the matter of which we’re continually tinkering with the instrument. Saying, “Well, that’s not quite an adjustment. We’ll make an adjustment now,” or, “There’s an improvement needed there.” So I don’t think they ever knew what their standard was, and that I think now has been pretty well documented. Their standard wasn’t at all standard. It was subject to shifts in calibration every time they touched it.
Did you know this back in ’81 when you were actually doing this comparison with their standard?
We had suspicions that if anything it would be better for the med office standardizing against than vice-versa. What we could establish was that our calibration agreed with theirs within ten percent or so. So we were happy that we weren’t a long way out, and we were happy that they weren’t a long way out.
So this is going to take a rough estimate check on the new instrument then?
Yes. Also it helped once we’d got the 31 and 123s side by side and they were both giving the same answers.
It was done at Howie?
This was done at Howie. Then again, you could be fairly confident that it wasn’t an instrumental artifact because we’ve got U.K. instrument 123 agreeing with 123 agreeing with 31 agreeing. In fact, really we’re suggesting that the measurements were okay and that any change in ozone that we’re seeing, we’re really in the atmosphere.
Now, how long did the 123 and 31 sit side by side down there?
For two weeks?
And 31 got shipped back?
Yes, because that was all the time that we had to release that. So the ship didn’t stay very long. Nowadays, the ships stay a lot longer and we could probably have them side by side for two months or three if we got the timing just right.
So you went down with it?
I went down with it. This is my first visit to Antarctic. This is all exciting and new. It’s still often exciting and new. I don’t think any of the trips that I’ve been on have been identical. We’re always doing something different. But it was really great because I was going to be there in the Antarctic actually doing ozone observations. I sort of knew the theory that I could do it, and so doing it in the Antarctic was really great. The hut that we had 31 in, it was pretty small, there was only room for one instrument inside it. So we’d have to have the other one operated the outside and we put a bit of cardboard down on the floor on the snow surface and had the instrument on its trolley sitting on the snow. That immediately gave us a few problems because we operating in about minus twelve Celsius. The instrument had been calibrated in terms of wavelengths, which in the U.K. I think the temperature range were 15 to 25, something like that. So we were way outside that calibration.
And the 31 in the hut was kept in that range?
31 is more or less constant temperature. Because it had been there for a long time we knew what the temperature coefficients for the [???] that’s essentially what we were looking at. But it became very clear that the temperature coefficient we were using, most of the D wavelengths were out, so we had to reestablish what temperature coefficient was.
That was for 123?
That’s 123. 123 was a new instrument, and the manufacturers couldn’t obtain natural cords for the prisms and the optics as a whole, so they fused cords and it turned out that the temperature coefficient for that fused cords is the opposite inside and twice the amount of the previously used natural cords.
I think Bob Grass had mentioned something about this.
Because we’d cool it down to such low temperature that actually gave us quite a good feel for what the temperature coefficient should have been. The way we do that is from the sky plots we scan the Q lever through the wavelength absorption and get your curve, which goes down and up and down again. So you know, again, if the first principle is, what the wavelengths should be on real sky, whereas, many organizations just use the mercury lamps, mercury and helium and all the other standard emission line lamps to do their calibrations. We prefer to use real ozone. So again, that’s something that is slightly different between us and many of the other organizations.
When you got 123 did you know it had the fuse quartz?
We knew it had fuse quartz.
But you didn’t know the coefficient was different?
We knew that the coefficient was different because the manufacturers established that at a workshop and they gave what they felt were the coefficients.
And they weren’t quite right?
They weren’t quite right, and they were particularly wrong in case of the D wavelengths.
Now, you identified this problem when you were doing calibration at the med office?
No, this wasn’t until we got down to Howie.
How did that actually emerge that there was the problem?
Because we were doing these sky tests, and the result that gave when the instrument was at minus twelve was vastly different from what you would get by extract lighting. So it was clear.
So the total ozone you were deriving?
No, the, it was in a sky — if I’ve got to any to hand... I know this is an actual one from 1982. This is ’92 when we also took 123. This is [???] in the U.K. But basically what we’re doing is we’re adjusting the position of the Q length versus the D wavelengths and looking at what dial reading you get. So as you come down you get a minimum up to a maximum and then down again. The actual setting for that particular pressure and temperature should be right about there. So we said a 106.7 is the middle. This is the actual ozone absorption for the D wavelength. It’s a fairly well defined and quite deep absorption light. It’s quite easy to plot. You measure what the temperature is, what the pressure is, and you also know from the manufacturers’ tables what you expect. So this is the Q value from the tables. In this particular occasion is point A to a degree different from what it ought to be. This is where we took two of the instruments at the same time. This one is number 73, and again, that’s 1.1 degrees. So the value drifts a bit with time. So what we did was one of these tasks where you’re just looking at [???] wave and you move the Q lever down and then back up again and you just scan through this absorption light. I’m doing this at a sort of a minus ten temperature. Howie was not just a little bit out it was sort of two degrees out, and that was sufficiently large. I was sure it was an experimental error, and so the only logical thing was a temperature coefficient.
So then you adjust —
Adjust the temperature code coefficient, and then you have to redo the ozone observations because the observations prior to that. Particularly the very low temperatures would be inaccurate.
You have this table when you take the measurements, you pull out the thermometer, do the temperature reading, and then check the chart on what the coefficients should be and that’s what you use?
That’s what we used then. I mean, nowadays it’s on the computer so you just type in the temperature and type in the pressure and it does all the calculation for you and it sets the Q lever, too, for its value. So you can’t do that.
So you identified that then when you first brought the instrument down?
So it was 123 that was actually down there doing the measurements in winter ’83, ’84, ’85?
So 123 wasn’t sent out — there was no additional instruments sent down there when you were getting these low values in those winter months?
What? So that we’d have two side by side at the same time? No, that was never the plan, because a Dobson is a Dobson and there’s no real point in having two side by side throughout the year if you’re convinced that they’ll record the same in the summer. What we would have done, if when we took it down they had given vastly different readings, I’m not quite sure. Fortunately, that didn’t arise. But if we put the two side by side and they’d been fifty percent different then we’d perhaps do some serious thinking as which one’s wrong. But as it was they agreed and the question didn’t arise.
The reason I was asking the question is because a number of things I’ve read say that the reason you brought the instrument down there was because you were getting low readings.
That was a little bit. We certainly wanted to confirm there was nothing wrong with the instrument, but this purchase had been planned irrespective of that to a large extent.
So when was it that you first started noticing the downward trend in the data?
In the late ’70s, really, but I suspect a lot of my present perception is colored by things that have gone on since then. Whether that’s an accurate recollection of what it really was at the time I’m not certain. But we had an open day at the survey where the public was going to come in.
What year was this?
I’m not certain of the exact year. That’s one of the things that’s faded from memory. It was certainly when I was working on the ozone data and also when there was lots of concern that aerosol cans would destroy the ozone. To some extent I thought, “Well, that’s a load of rubbish. How on Earth can aerosol cans possibly destroy the ozone layer?” We’ve got data in the Antarctic and I’m working on this year’s data. We’ve got data from ten-ish years previously. I’ll do a comparison of the two and that will reassure everybody. The problem was that the October values in particular were quite a lot lower than they had been earlier. We’d been experimenting with extending the observations season by changing from CC dashed to CD observations because you could do those at high [???] values. So the observations weren’t a hundred percent comparable with what had been done earlier. So it could have been an artifact, but it certainly seemed to me that maybe something was happening. And also, occasionally the guys on base would say that the ozone values look a bit on the low side.
So you think this is around the late ‘70s?
This is certainly the late ‘70s. Then what really happened then is that the process in the radiation data was coming to a close.
... just doing a fairly rough calculation of what it all was. I think I also went to Joe to say, “Look, this is really exciting. The values are low in October.” He said, “I know. October values don’t really count because the winter time ozone values were all really indeterminate. His view was that you’ve got the spring warming that just swept out and in. It just affected the winter values depending how much spring warming you got, and that wasn’t predictable. So October values weren’t predictable. You might get two high years and then two low years. I think his view was also colored by the Hamilton paper, which had looked at the day-to-day variation in ozone amount and identified the southern autumn as being the best time for detecting changes in ozone because the day-to-day variation is least in the autumn. So February, March time, maybe even January was the time that we’d see any changes was October, November, because the timing of the warming was quite variable. But values could be high or low.
This is what Brian Gardner had mentioned that in the ‘70s when Roland and Molina — all these predictions that fluorocarbons would be bringing down the ozone. That’s when you looked for it initially, was February to March.
Yes. I didn’t really have any preconceptions as to what should be high or what should be low. I was just looking at it from the point of view of an observer who had no theoretical background at all. But I was a physicist, so I was sort of looking at the data. It certainly struck me that October seemed to be low in the years that I was looking at compared to the previous data. At that point it would be fair to say there was nothing systematic about it. Just one year, and as Joey [?] said, “One swallow doesn’t make a summer,” to use an English proverb. So it sort of got left at that for a while. Then the following year October values were, again, low. He still wasn’t convinced. But again, in the background we’ll still be doing all the processing of the 1972 through the current batch. And at that point I began to focus on the Howie data rather more and started getting through it as much as I could. And was then able to demonstrate pretty clearly that it was a systematic event, which each October we’re getting lower and lower values.
You could tell it wasn’t a calibration problem or anything?
I’d gone through all the calibrations and was reasonably happy that they might only be accurate at the ten percent level, but this effect was above any reasonable noise.
And you still think this was prior to you going down with 123?
No, this is after.
So this would have been?
’82 or ’83, something like that. This is my original discovery paper, if you like. This is where we’re going.
You wrote this up?
This I wrote up and used it to convince Joe and Brian that it was happening. And I submitted this to both Brian and Joe and Michael Wycoft [?], who was the head of the division at that time. Michael came back with all sorts of changes and Joe went away smoked his pipe and thought about it. But basically, the key thing was the values did seem to be dropping.
How did you determine this down [???] ?
Just [???] squares. That was a January launch I thought was going down, but maybe not. But this was the October one, which is pretty clearly going down.
So this would have been in the fall of?
This started in November of ’84. That was what kicked everything finally into action to get the Nature paper sorted out. Prior to that —
Can I get a copy of that?
Yes. I’d actually written to Harry Broxom [?] at Wallace, and was 5.2 or something at the [???] level.
And this was about the low values?
This was way before we published in Nature, saying that the values are definitely low. I thought, “Well, is it El [???], which had gone off rather at that time.” We’d actually asked them here, “Is it confirmed by such a [???] ?”
And they pretty much wouldn’t have known.
Well, they should have done because we were doing comparison observations for them. We were doing ozone observations to coincide with sun light and over-passes specifically to calibrate satellites. So we were doing these special observations for them, and we thought they really ought to be able to tell us whether the satellites were agreeing. I guess we were lucky that I only had one reply to that from this guy and that group is no longer involved. So if they’d actually bothered looking at that point they could have beaten us into the [???].
As I’m finding out, people down the hall didn’t know what was going on in the ozone processing.
This is very perplexing, because as the story goes, your paper comes out, NASA went back and reprocessed it and found the hole. But as I found out, they actually identified it in July of ’84. But the problem I’m having is that it shows up in the previous years, but they didn’t identify it prior to the summer of ’84 and I’m having a little bit of difficulty understanding why they didn’t.
I think probably it’s just having somebody look at the data, because very often it’s how you are looking at observations that govern what you see. I was just looking at it almost with a blank slate and I’ve got no idea what to expect. And possibly having the long gap where we hadn’t been doing anything was a help because it was unclear that the October I was looking at were different from the earlier ones.
The long gap in the process of the data?
Yes. Whether it would have been as obvious. It ought to be as obvious if we’d been keeping up with things, but especially if we were doing it operationally as process data you wouldn’t look back quite as much. You’d only look at what you’ve got now.
So mostly finding out that’s what was going to maybe go with the Nowa [?] people.
So largely said, that’s what you do with operation meteorology. It’s the observations that they’re making now that matter because that’s what you want to use in the forecast. It’s all real time stuff. What it was ten years ago and whether this year’s is different than ten years ago isn’t really relevant. And I think that’s one of the problems with the climate change debate. That most of the meteorological measurements are not made with the use of detecting climate change. They’re there for forecasting purposes so that you can get a good picture of what’s going on now. And if we were doing it for climatic purposes we do it in a very different way. I think that’s probably true of the ozone instruments. That the interest was what’s going on now rather than is it changing.
You were looking at the autumn values. You weren’t really seeing anything there. Maybe a little bit, but the October values where you really saw it. You said you were focusing on the Howie Bay data. This shows up in the Faraday data, too.
It does, but not as clear.
Now, did you see it in the Faraday data before?
At about the same time, yes.
It just wasn’t definitive enough?
It’s not so clear cut. What I wanted to do was show here this is concrete and highly significant. The Faraday stuff was significant, but less so than Howie, which is why it’s always focused on Howie.
Is the problem with the Faraday station that it can be inside the hole? Outside the variability is much greater there?
Yes. And also you can get very large changes from even one day to the next. It can change by fifty percent or more from one day to the next at Faraday. Whereas at Howie the day-to-day variation is very, very much smaller than that.
Now, prior to the early ‘80s and the ‘70s when we were talking about the CFCs and all of that, do you recall having conversations with Brian Gardner, Joe Farman, about looking for this stuff from the data? Or do you do this on your own?
I think it was really only sort of outlines that the autumn was the best time to look at changes. But it would be sort of nice to find something in our data, but we thought it would be fifty or a hundred years in the future before anything significant was there. I think it was a vague conversation about the big solar proton event in ’79. I’m not sure of the exact date. Big solar proton event, which somebody said it affected the ozone layer. We couldn’t see anything in our data, but then it was only supposed to affect the very top upper most part of the atmosphere anyway. That’s, as far as I can recall, the most we talked about it.
So you’re looking at this stuff then was pretty much you were involved in processing all this data.
My job was to get the data processed and make the final report as accurate as we could.
It started emerging in your consciousness. People saying values getting lower, you’re noticing it —
Yes. And sort of putting two and two together, and well, it looks like there is something here. So the next step is can I actually demonstrate that it’s significant and [???] document. Yeah, I could. And it was really that, I think, that convinced Brian and Joe, that yes. And they then thought, or particularly Joe felt, that it’s no good just publishing observations. You’ve got to explain what’s going on as well. So he then went and looked at his model with a view to trying to use that to explain what was going on.
That’s one of the things that held the NASA people up. Well, in the paper it says, “To interpret the difference as a seasonal instrumental effect would be inconsistent with the results of routine checks using the standard lamps.” Can you comment on that? And also the fact too that you’re having a seasonal effect here. Just prima-facie seems to suggest that it’s not an instrumental artifact. So could you speak to how you could think that a seasonal effect was an instrumental artifact?
It could be related to temperature, for example. But in general it will be somewhat colder in the atmosphere in springtime compared to the autumn just because of the seasonal lag in solar effects.
In that case wouldn’t it be symmetric at the beginning of the spring, the end of the autumn when it starts to get cold again?
Yeah, so you might expect to see some symmetry there, but it could equally be an environmental if the temperature instrument was changing from spring to autumn. So that could introduce a bias or low —
So it’s changing from cold to warm as opposed to warm to cold.
Although, the fact that the instrument is in a thermostatic hut and there’s no evidence for it changing. But that was a possibility, and those sorts of things should be shown in that monthly standard lab tests that we do, which are looking at the long term calibration of the instrument. If the processes weren’t [???] you’d expect to see it fairly clearly. There was obviously no trace of seasonal effects in that. But I think that needed to go into the paper saying that we’d looked at all that.
Blocked that possible interpretation. Now, what about the fact that the instruments cleaned over the winter? Was there a thought that maybe there was something —
That it could be getting dirty. That certainly could be a possibility except that if it was, again, you’d expect to see a jump in the standard lamp readings.
And that wasn’t showing?
When you do get a jump generally what happens is that all three wave lengths seem to jump together. So because largely we’re basing this on AD or CD observations, those effects get taken out of the equation. Again, that was one reason for wanting to go away from the CC dash type observation because that’s just using single wavelength. So it can swing as individual wavelength calibrations change. But using paths that [inaudible] together and you’re subtracting A from B. That just cancels out, which is one of the really neat things about the way the instrument works. But by and large, everything that you can do wrong cancels out so that really demonstrates the brilliant design of it.
Are there any other potential problems that were considered that might be going on with the instrument?
Observers. And that can play a part, but again, we change them every two years. Seems to be independent of the actual observer. The orientation of the instrument, it had always been aligned correctly with the sun when you’re doing the observations, and maybe in the spring they weren’t doing that and in the autumn they were. So that can release a slight variation in measurement.
How was that ruled out?
Largely by experiment and demonstrating that it only makes a few Dobson units difference. So that’s well in the noise.
Something like that, was that done here?
No, it was done in the Antarctic.
So you actually asked them to locate the instrument, see if that produced any —
We did sort of start writing it for a little bit that it is important where there’s any observations to make sure it’s aligned. There was a procedure for the direct sun observations, but usually it sort of left [inaudible] observations whether you need to align the instrument or not.
You wrote this paper up November 2, 1984. You submitted to them and things switching to publication. After Joe Farman and Brian Gardner read this, what sorts of conversations did you have with them?
I would say almost deafening silence, if you like. They really sort of just went away and dotted the I’s and crossed the T’s on everything, and it was really getting me to check the data one more time. Is it all a hundred percent? And then to draw up the graphs demonstrating it. I think I’d used the lowest springtime value and I also used the lowest October value. So that’s a little slight difference in emphasis.
They were becoming convinced after you read this. They didn’t just automatically, “Oh, yes. This is real.” Or did Joe Farman need a little bit more convincing?
I think Brian had to keep on at Joe to get him to think about all the possibilities and also to get it into print. Joe would have gone on forever making sure that everything had been thought about and there weren’t any possible errors, whereas I think Brian saw the need for urgency and get it out as soon as we can.
Do you recall when the paper was actually sent out?
I’ve gotten some dates from —
Brian. Brian actually put it in.
Kind of triangulate on things because he’s pretty sure, but not one hundred percent sure. Now, the paper goes off to Nature. Were you talking about this to other people in Cambridge at the time? Were you getting feedback from people?
Not in my case. We were sort of talking about it with the Antarctic observers because they were the ones making the observations. I think they were surprised that it took so long to get it to Nature.
So in between the time it went to Nature and the time it actually came out was there much talk around that you’re aware of about this?
I really don’t recall anything particular. It was really sort of feedback from the referees that wanted a few points clearing up, but that it was generally positive and they liked it.
You didn’t get involved at all in the discussions of the explanation of this effect?
No, I was the observer, and here are the facts. You’ve got to explain it.
How do you feel about the ability to say that there’s actually something there? A real effect when you haven’t dealt with all these other theoretical possibilities that Joe Farman wanted to look at?
I think largely from knowing how the observers operate and the way the Dobson’s functions it’s very, very difficult to get something that happens in October that doesn’t happen in March. At that time, there wasn’t any trace of any changes in March. So it really had to be a seasonal thing. It seemed fairly clear to us that it involved CFCs. Although the graph that we published in Nature was an artifact, if you like. We carefully adjusted the axis to match, which probably we shouldn’t have [???].
Yes. And also possibly in reverse for CFCs increasing matched the ozone decreasing. That really did seem to be the only plausible change in the atmosphere. At that time I don’t think we thought about [???] in the slightest. We knew we had them and to some extent we perhaps — I’m not sure whether we actually thought they were becoming more common at that time or not. I think probably that came a little bit later.
But so far as the data you had from the Antarctic, it was showing that you have this drop in the ozone, but there was no real significant changes in the dynamics of the polar vortex? Or could you tell that from your data?
Not from the ozone data by itself. Temperature data really didn’t show much in the way of changes, but it’s pretty messy. And also we changed [???] systems, which you couldn’t then be certain whether it was a change in the atmosphere or a change in the instrument. But again, it’s something that’s good about the Dobson measurements that were being used. The same instrument since 1956 with the same techniques, but again, we made it pretty clear that it really couldn’t be an instrumental artifact. Which is perhaps what the space base people couldn’t be so certain about, because they keep launching the satellite and they’re using the latest technique to do the measurements. So one of the changes that they were seeing due to a new satellite to the atmosphere. And I think also that the space base people think or it must be due to the sun to some exchange and we had satellites up to six years and the solar cycle is going down so the ozone is going down. And they appeared not to be willing to consider a longer historical record, which went back over the sort of thirty odd years at that point.
Are you talking about “they” as people at NASA, or just the scientific community in general?
I’m not sure if it’s NASA or the scientific community in general, but certainly the space base community seemed to want the sun as the mechanism.
Like the solar proton events? Things like that?
Yes. We’re saying that as part of solar cycle, clear correlation. Because they only have the data for half a solar cycle. So I think our view was that’s not very physical and you can’t draw any conclusions about correlations on half a solar cycle. There are people still doing that. They’ve got half a solar cycle or two solar cycles, but with all this wonderful correlation.
So you think you data could strain the situation, especially to include that in interpretation?
You’ve heard of Shigeru Chubachi, right?
Japanese polar researcher who reported low values in ’84?
What are your feelings about his reporting those low values and what he was up to as opposed to the results you got at Howie Bay?
He couldn’t draw anything conclusive out of it because they didn’t have a continuous record. I think he’s got a few spot measurements.
They had been taking measurements at Showa us since 1966, right?
Yes, but nothing like ours, as long as ours. And also Showa is not as well placed as Howie was. Howie tends to be closer to the ozone hole, so we tend to get lower values.
So the low values, they were getting — could have been due more to planetary waves and things like that?
No, I did actually look at most of the over Antarctic solutions from the red book.
Around the time you were looking at your data?
Yes, because we wanted to find out whether it was just us or everybody else. I think most of the others showed reasonably plausible for decreases, but nothing like as conclusive as the Howie situation. I think I wasn’t aware of that paper until long after the fact, but I was aware of the Japanese data and the red book.
Being a bit low.
And that was low as well.
You checked and they did have profile data from Showa. Did you look at that at all or were you just not concerned since you were only doing total ozone?
We weren’t particularly concerned about it. We had our own profile data from Faraday. [???] have been Howie. That didn’t really show up anything terribly clear cut.
So you’re saying then that these other stations were showing some decreases, but it wasn’t —
Only because their record was patchy or it doesn’t begin early enough at the South Pole. They can’t make out any measurements until most of the ozone hole is gone just because of solar elevation. Or the stations on the other side are more affected by a spring warming at that sort of times. So they don’t see it. So it’s really very Atlantic side of Antarctica that is well placed for seeing changes.
Was there anything else in the processing of the data that caused potential problems for interpretation?
I don’t think so. Once I got the computer programs roughly sorted out, then, provided I’d done the programming right, that was it. And I think Joe and Brian worked through some of the results by hand just to make sure that there wasn’t anything seriously wrong.
Brian was saying that they had looked at the direct sun observations because it was less potential for —
Direct sun observations are absolute, if you like. [???] are all by relation to the direct sun. So if you get a direct sun observation and just look at the direct sun as it shines on something, then you can be 99.9 percent sure that it’s real.
So it’s just less chance for things to go wrong with direct sun light involved?
And also the accuracy of the direct sun is better. A well done AD direct sun should give you ozone values to [???] percent. If the instrument calibration is correct.
Now, had you thought about doing that when you were processing the data?
Extracting out just the direct suns?
I was more working on daily means because I thought if the daily mean gives you something, then —