Oliver Smithies:

This is the last two pages of book 1, they have rather an interesting history. It’s really concerned with loot from WWII, oddly enough, that when at the end of WWII various items that were in Germany were basically looted by both the US and Great Britain. What I’m talking about here for the last two pages is a microbalance, made by Sartorius, a very famous balance maker from Germany. In fact, our current balance in North Carolina is a Sartorius, beautiful balances, and quite impressive machines. Anyway, Sandy Ogston in some way or another got ahold of a Sartorius microbalance, that’s a balance able to weigh micrograms, from Germany, but it had never been adjusted properly. And there are various knife edges in a balance, three at least, one which is the main knife edge on which the balance beam rocks, and then there are two knife edges supporting the two pans. In this type of balance, one side had a weight and the other side had the object you were measuring, and the deflections were used to find how much the right hand side differed from the left. But this required that the knife edges be adjusted in a very special way, because the center of gravity of the balance has to be below the knife edge that pivots the beam of the balance. And the distance the center of gravity is below that knife edge determines the sensitivity of the balance. So obviously it has to be balanced rather carefully. And the other two knife edges also have to be adjusted. And I was given this balance to try to calibrate it, and I might say I was never able to calibrate it, because I didn’t know which order to make the adjustment. So anyway, on the fifth of whatever month it is we’re in – the fifth of March, I began to try to try to adjust and calibrate this microbalance. So the first thing I did is I leveled it with a partition level, focused the scale and fixed to zero. And I had a five milligram weight, which I could put on the left hand pan, it was a physical weight, and adjusted to zero, etc. with what was called a rider. And the rider was adjusted to be .02 milligrams on the left, and the reading came out to 20.4 micrograms. The temperature was 19.6. And I adjusted to the new variance and I got a mean of 20.5. The zero was adjusted etc. and I did it again, and then got 21.2 and decided to clean the balance and knife edges and did this and repeated it, and got various readings. But they were not stable, and so I dismantled everything and cleaned it more extensively, and reassembled adjusting the positions of the knife edge relative to the guides, etc. etc. and tried again. And this time I put a two gram weight in the two sides, and couldn’t get it to come correctly. I got 19.9 grams in one side and 18.8 if it was the other. I think it’s what its meaning (?). I cleaned these knife edges, etc., and took everything into the constant temperature room. And I even have a graph of how much they were deviating from a straight line, that I don’t quite understand. Anyways that’s what I was doing at the end of Book 1.

So Book 2 continues with this attempt to adjust the balance. Now I’m talking about the 13th of May, 1949. I don’t know how that jump in time occurred, but at least it says 13-5-49. The scale arm was loosened and moved so as to bring the beam etc. to equilibrium and left hand side, right hand side, zero difference. 19.2 on the left hand side and 18.2 on the right, etc. etc. This sends it slightly over to the left, but the effect is very small in the sense that everything is wrong, so I had to try raising the center of gravity. This altered the sides unequally. So I’m trying again, another page of doing it, with a two gram load, and I’ve got 10 on one side and 10 on the other, plus or minus less than .05. I seem barely happy with it. Then I tried with the sensitivity versus the load, .1 mg on the left hand side, to simulate the usual weighting conditions, 1, 2, 5, 10 grams, etc., and how much did it come, and it wasn’t really very good because the sensitivity varied with load, so that at 1 gram I got 10 micrograms, but with 7 grams on either side I got 9.4. So it was not able to have the same sensitivity with loads. I decided the pan edges must be below the central edge, and tried messing around with that and noticed that the knife edges were not quite straight, and so on. Messing around with it. Test on the balance shows considerable shift with load. That page and other page, sensitivity is 7.9, raised the center of gravity and retested. Here I am now, 24th of May and still messing around with it, not working very happily. Moving the center of gravity, the balance arm, calibration of the weights. I was using a national physical laboratory weight as a tare with weights on either side, etc. Try weighting different time, I’m still working on it, hopeless here, I’ve got a note on 2S2G – hopeless. Removed air dampers, there were some air dampers and things. Here I’ve got pages and pages of this stuff. The last experiment suggested temperature variation causes irregularity. So I’m trying to make this thing work. Gloves used, two doors of the outer cage kept opening. But not the glass door, etc. Trying again to get this thing to work. A note: discontinue for the time being. Then here I am the 31st of May, still trying with this thing. The weight of the protein, and the weighing box, I’m trying to weigh something vacuum dried, etc. transferred to the weighing bottle. I think this actually must have been on a different balance, I think I must have given up the experiments with the Sartorius. Discontinue for the time being, then there’s a blank page.

So the next page has nothing to do with the balance. It’s just me trying to measure the weight of my freeze dried protein. It’s a little odd, I had 4.55 milligrams of the protein recrystallized to be attempted, dissolved in water in the electro dialyzer, dialyzed against de-gassed water, so I’m back to working with protein solutions, trying to get dry lactoglobulin. So here we are making .1 mole of sodium chloride in which to dissolve it. That I need to have 33115 grams. That’s sort of hopeful precision. Five decimal places. Sodium chloride has molarity of 0.10004, so obviously ?. Everything’s written to so many decimal places. 24 milligrams were taken, dissolved in water, in one mil of .02 moles of sodium chloride, etc. etc., dialyzed, small sample was examined, crystallized, small but perfect crystal. Typical Smithies. Here we are with an ultracentrifuge run, 718. Sandy Ogston had one of the only three Svedberg ultracentrifuges in the department of biochemistry at Oxford, which was principally operated by Rupert Cecil. I tried to avoid the machine because it was such a huge complicated thing that I didn’t really like it. But anyway here I am with an ultracentrifuge run, first one I’ve recorded in Oxford. And I measured the sedimentation constant, it’s 2.80 times 10 to the minus 13. And the plate looked very symmetrical peak. Plates showed complete homogeneity, is what I said, at least with this test. Somewhere here I’m worried about determining whether loss of concentration occurs. Protein crystallizes were centrifuged and washed and dried. Some loss of very fine crystal occurred because they bumped out of the ? in this stream of water molecules. This did not seem to occur until the vacuum had been broken to adjust to join (?). So do not therefore break the vacuum once started unless essential. Trying to dry my protein without losing anything. Anyone who has done freeze drying of proteins knows what I mean. It’s very easy for bits of protein drying to follow the stream of water molecules and disappear out of the drying part. Measuring refractive indexes. Refractometry of .58, 0% lactoglobulin. Now the 18th of June, 1949, still not very clean and leaked, dissembled, cleaned it, and reassembled it. Similar things going on all the time. Trying to get good clean lactoglobulin. Again, here’s the diffusion constant. With the sedimentation. Indications of a little peak that was faster than the main component. So it might be really a mixture of two proteins, one smaller than the other. Calibrating the built refractometer again. Using potassium chloride, which is .19999 molar. How many decimal points does Smithies use? Tare: 1.43252 with a correction to 51, plus or minus 3, plus 3 minus 4, in that decimal point. So I’m talking about measurements to four decimal places. The potassium chloride solution, I end up by saying it’s .9406 molar. It determines the cell constant. Repeating heating experiments. Here’s where I made a comment somewhere earlier you could heat lactoglobulin crystals as they dry – here they were heated for 4 hours at 105 degrees centigrade. Though initially 15-20 minutes at 140 degrees centigrade. It didn’t do any harm to the protein as far as its ability to go into solution was concerned. The final solution, very slightly opalescent, no precipitation, the solution of the solid was very ready. So it’s extremely resistant to being heated when it’s dry. More ultracentrifuge experiments.

And then another preparation as usual of lactoglobulin, you have to make the stuff time and time again, here we’re talking the 7th of June, 1949, two gallons of raw milk, etc. etc. This time I used a better grade of ammonium sulfate. What sort of filter paper do you use? In those days there were many types of filter papers, and I liked 54 papers. They were very hard and very thick, and they didn’t lose little bits of filter paper into your solution. Very thin papers, very hard, 54. I still remember them. At one time I had a huge collection of different filter papers.

I have a note here, attempt to use somebody’s electrodes for pH measurements, with the comment “no use, these electrodes are done”. Glass electrodes, this is a glass pH reader. I don’t know who DC is. Oh, must probably be RC, it’s probably Rupert, Rupert Cecil’s electrodes. Yes, here, the next page it is new electrodes made with RC’s blanks. So worrying about pH meters now. Repeat on the microbiology’s pH meter set up with borate which is a constant pH of 9.23. Later on I used that quite extensively in my work only a year ago. Borax when dissolved in water always gives a pH of 9.23 or 9.13, whichever. I think 9.23. Standard pH. Here I’m trying on the 30th of august, 1949, to calibrate the apparatus used for measuring surface tension, the capillary rise apparatus. Trying different editions to alter the surface tension. I tried adding teepol. To ensure accuracy all the former tubes and the u-tube were boiled in sodium hydroxide and then in distilled water, left overnight in distilled water. Washed out and started again. Three drops of teepol were added. Various measurements. ½ cc teepol to one liter, repeat tomorrow. Attempt to ensure that the meniscus was always at the same place in the apparatus. So pages and pages of worrying about surface tension and whether I could get rid of it with teepol. Still more pages of teepol and distilled water. Repeat. Pages and pages. Move to the classroom, organic solvent vapors in the lab. I wasn’t happy with organic vapors in the lab so I moved to the classroom. A new lot of distilled water, even. Distilled water and teepol, more things. Acid cleaned everything, washed well with distilled water and started again. Each reading to be taken at more than 5 minutes apart after steady reading to try to avoid some effects of time. Always worried about super precision. There’s a diagram of the apparatus which I don’t remember what it was, it has a funny shape, but I don’t remember how it worked. Cleaned and fresh start. Quite clear that a time effect responsible for the irregularity. A rough curve is obtained if an adequate time is given greater than 30 minutes with a low concentration of teepol. So I’m using teepol just to simulate protein, to see if I can measure the surface tension correctly. And I have here a comment, triethanolamine oleate seems best available as a standard. Back to recrystallizing lactoglobulin on 9th of July. I guess it’s not, it the 7th of October. Then I here I am, last surface tension of lactoglobulin solution being measured. Still measuring surface tension at the end of Book 2.

Book 3 is beginning on the 8th of September, 1949. That’s where all the surface tension experiments continue, so I’m measuring the surface tension of lactoglobulin. Decided to abandon this method, which was a method using the previous cells and began to think about the method that I eventually used, which is to make a dummy osmometer, with a tube selected by capillary tests to give the same rise as in the apparatus itself and then to have a buffer with a rubber latex sac, as I was describing these things from the thesis. This is when I first began to think about using a rubber latex sac to be placed and collapsed. Then read the surface effects directly in the apparatus. So all those earlier measurements that were made in a special apparatus were going to be replaced by measurements made on a dummy osmometer. Check in the diameter of the tubes to make sure that there was not a difference in pressure due to the diameters of the tubes. I have a note “before the measurements, all apparatus was cleaned with nitric acid and ethanol.” That’s that method that I’ve described before which is very dangerous. And then with water and alkaline potassium hydroxide, even alkaline potassium hydroxide is very dangerous, water and distilled water to follow the earlier practice. Everything had to be impeccably clean.

Here’s a place I think I mentioned before, somehow. Here we are – the former had been left coated with silicon. It was cleaned by boiling in aqua regia for several hours, then potassium hydroxide and ethanol. We must have found that in the thesis, because here is the page in which it was being done. That was commented on in the thesis. The comment here that after this work on the 13th of September, 1949, saying that the u-tube and either former can be used to indicate capillary rises in the osmometer to better than .01 centimeters. A tenth of a – well let’s see how many is that? 0.1 of a millimeter, which is the level I decided I wanted to do. My initial attempt was with surgical rubber finger stalls. They were too thick for the sac. They were made by British Rubber Products Research Association. So I decided to make the membranes myself. More lactoglobulin. Precipitation as usual, crystallization, re-crystallization in the meantime. Two methods of trying to prevent losing crystals of protein when it was being dried. Neither very successful. So I tried a different one. I have a note here, 9:25 pm, freeze drying stopped for night and flask rubber stopped after emitting hair through sodium hydroxide and liquid air traps. So I let air back into the system that had been fully dried. The powder says no jumping, very dry in appearance. Trying to get accurate level by weight of lactoglobulin.

Check of rubber sacs, the British Rubber Association sent me two latex sacs, wash them well, etc. etc. Found a difference and said they’re much too thick, she’ll have to wait for thinner ones. So we got latex again from the rubber association, and made the sacs myself. Test of the rubber sacks. Half a dozen were made. The formers are numbered 1-4 and then were tested. What the difference was on the two sides of the dummy osmometer, quite small differences. Different formers, different u-tubes. Here’s a note:  it’s clear that the formers 1 and 2 are used together as osmometer and surface tension compensator. Errors will be very small, less than .005 cm. Testing these sacks. I made one of ½ the thickness, tried to make it yet thinner. The difference in pressure on the two sides of the collapsed one were really very nice, .002 to .004 cm. and I may mean for .001 cm in precision, .1 millimeters. So, I’m not really very worried about them. Removed the sack and retested, etc. etc. sacks made on Friday used on Monday. Nearly final conclusion is that these thin sacks do not introduce any appreciable constraint on setting up of levels, provided that they visibly collapse. Addition of a little more buffer should not alter the level difference when the sack is properly collapsed. This should give adequate tests, if the collapse not visibly accurate. I was happy with these little collapsed rubber johnnies. Worried about the apparatus being cleaned. Obviously uncleaned. Nitrous acid and ethanol again, that disastrous mixture, to get things clean. Further check on the rubber sacks performed by measuring the level with the sac and the difference is only .001, which is a hundredth of a millimeter. And it says here that a reasonable but not excessive cleanliness was used. All of the formers within the limit of .010 centimeters of each other, so I’m happy with them. Now I’m back to osmotic pressure trial, osmomoter itself, this is 16th of November 1949, made up fairly accurate .1 molar sodium chloride, 5.85 grams needed. Chlorobenzene droplet tried again, using the sample – that’s a droplet that goes in the horizontal tube of the osmometer. I have a note here that on Friday afternoon the droplet was placed in the tube and it was slightly deformed by the tube so it was completely blocking the tube, as it should. But on Monday, it was now no longer blocking the tube. Friday afternoon the droplet was placed in the tube, on Sunday it was still alright, Monday morning it was no longer blocking the tube and was visibly smaller, and later on it got smaller still. So some solution is occurring. The same two times re-distilled. Also chlorotoluene was tried, its solubility is less, and the density is 1.08. So I’m trying also chlorotoluene instead of chlorobenzene. And I remember that in fact was what I used finally. Droplet very little smaller overnight. Also chlorotoluene is therefore suitable, continue checking, mixed in the droplet to move for at least one minute. And I had a comment here, leakage BOKFO but ok for Oliver.

Thermostat overheated, change to a 50 watt bulb instead of evidently a 50 or bigger. Remade the joints. Set up on the 24th of November about 20 centimeters of water overnight, mixed in the droplet is now correct, and tried several timings for each point. Back over the page one way and the other, learning to do multiple measurements. And did outer centrifuge run for somebody here, D. Hamer. Don’t remember who D. Hamer was. We set up the drop again, got better results by Monday. Those readings were done on the 24th and then on the 28th the drop was still ok, so o.cl.tol. (?) is evidently quite suitable. And then I have measurements plus or minus .02 centimeters, which is five times better than I wanted to get as my accuracy, so I was pleased with it. Particularly fast membrane, nearly two times faster than usual, plus or minus .001 centimeters. That’s the .1 millimeter that I was aiming for. Plus or minus .015. Dirt in the u-tube, cleaned and continued, dirt in the u-tube. Altered my precision. Very much better. Plus or minus less than .001. She’ll adopt this method. 11.16 plus or minus .01. That’s centimeters. So that’s within a tenth of a millimeter that I’d set for myself. Calibration of the manometer for surface tension effects, and the apparatus starting on the 12th of December, 1949. And several pages of measuring that. And a summary of how to set the apparatus up. Method of setting up or method I was going to use was set up the manometer, u-tube, and osmometer, cleaned overnight, acid ethanol, then just before use, potassium hydroxide and ethanol, and then distilled water, etc. etc. A whole list of things, how to do it in a different order, doing things (?). I was a bit worried about dirt getting into the system. Here’s a sad one, the u-tube was broken, repaired with the same tubing. Thank you to the glass blowers. The new u-tube shows levels accurate enough for all formers. The formers were the things used to make either the rubber sacks or the semi-permeable membranes. A comment here on the rubber sack. The degree of collapse is not crucial, but it’s better to be more rather than less. None of them gave much of an error. The rubber sacks adequate to less than plus or minus 005. Now here I am on what’s called pre-equilibrium. First attempt at using the dried out membranes and dialyzed lactoglobulin showed considerable time for it to come to equilibrium. Part of the cause might have been that the dialysis sack was not stretched, and may have been part of the cause, so I tried to stretch it a bit. Noticed the leakage in the rubber sack, and therefore tested it before using. So over stretching them is not a good idea, it makes them leak. Continuing rubber sacks, now I’m testing on the tenth of February. Decided to try heating the sacks before removing them from the formers. Stretched them. It didn’t seem to make any difference. Obviously concerned about these sacks still. Tested the collapsibility of the latex. And the sacks have been steamed for ten minutes on the 15th of February. Degree of collapse increased considerably. If they get complete, it’s not good, so you have to have a sack that’s neither over-extended nor too compressed, here’s a little diagram of what you shouldn’t have. Test of collapsibility. It looks like there’s a new lot of these sacks gave none were leak proof, they had to be double coated to prevent them from leaking. Test of collapsibility, still doing it. And that’s the end of this book, still working on the collapsibility of the things to measure surface tension. So there’s so much work in trying to get something which is only a minor correction. Amazing.

So calibration of the u-tube, nitric acid and ethanol cleaned, potassium and ethanol cleaned, and then distilled water. We’re starting on the 23rd of March, 1950. I was unhappy about what was going because it turned out that the tubes were not uniformly parallel, they could be slightly tapered, so I had to get a new type of tubing called veridia tubing, which you could buy where the diameter of the tubing was guaranteed to be uniform all the way up the tube. You could buy that and make the apparatus from that. And a standard flask, I don’t know what I was trying to do. Oh, this was calibrating the flasks used for making solutions. So I found that it wasn’t 500 milliliters, it was 499.5. An error of one part in a thousand. This first can be taken as 499.5 ccs plus or minus .1 ccs. Completely crazy. I don’t know where this impossible focusing on precision came from. Later on I learned to do what I called Saturday morning experiments, where you don’t have to measure anything. You have to keep notes, but you can use a spatula for measuring and take a pinch of that and a pinch of that, and do a crazy experiment on Saturday morning because you’re allowed to be crazy. Anyway here I am making lactoglobulin again, and the electro dialyzer, trying to get it to have a better flow pattern of water so that it gets rid of the salt more quickly. So numerous attempts, they’re mentioned in the thesis, of trying to get the flow pattern in the electro dialyzer better. Not very important, but still pages of doing this. And a different way of trying to prevent losing the protein when freeze drying. Protein solution is being made, pre-dialysis technique. Used for .7104 molal sodium chloride. I don’t know why I was wanting .0710 molals of sodium chloride. But a different method of dialysis against it. With the rotary re-dialyzer Sandy Ogston had made that went backwards and forwards, or round and round, round to the left, round to the right. Calibration of the thermometer, now I’m calibrating the thermometer, the total immersion thermometer, vertically, minus 5 degrees centigrade to 50 degrees centigrade. Centigrade in 0.1 degrees centigrade division. It was calibrated thermometer number EN80740. I’m trying to find out what the correct temperature was versus the reading. I don’t know where I got the constant temperature from. Must have been calibrating my thermometer against this military standard thermometer. So the military standard thermometer I took as being correct. The correct temperature is 40 degrees, and the reading of my thermometer was 40.01. 25.06 and at 20 degrees is was 20 degrees. So the calibration of this thermometer was good to .02 degrees centigrade. So measured with a national physics laboratory standardized thermometer 2476 which is calibrated at 40 degrees, 25 degrees and 20 degrees, which had been corrected. So I was using a national standard thermometer to calibrate my thermometers.

Calibrating the refractometer, now using potassium chloride solutions which have known refractive index. Everything was to try to get more and more precision out of the thing. More refractometer measurements. My former was broken and repaired. Trying to dry the proteins more effectively without losing it, made a little apparatus. Calibration of the refractometer again. Somebody else’s freeze dryer I tried. Weighing bottle here, the hope was to make 0.900 molal sodium chloride. High precision again protein solution. The apparatus which was finally reduced to make the solutions that I measured, my final measurements were sacks containing 2%, 4%, 6%, 8%, 10% lactoglobulin for dialysis. I refer to this in the thesis, so some of this is in the thesis. Some of the thesis information is in this book. Manometer tests with this new tubing, it turned out that the tubing was what I called semi-matte internally, and the hysteresis is very marred, that’s to say going upwards and going downwards it remembered which way, water alone doesn’t improve this much, water and teepol removes the hysteresis, but it’s still present. It’s clear that mercury isn’t suitable for smaller osmotic pressures.

I’m calibrating the balance now. It says “rough calibration of large weights assume that the Sartorius weights are accurate enough, other kilogram weights are adjusted to one kilo. One needed 100 milligrams to correct the 1000 molecular weight standard and the other needed 200 milligrams so that I’m correcting the balance weights now, calibration of balance weights.

Some lactoglobulin 14, X14, must have been the 14th lot. So it goes, so it goes.

Electrophoresis, ?, leading to uncertainty in the charge. I didn’t remember doing any electrophoresis experiments in Oxford, but I must have done. I’m back to collapsing rubber sacs and membranes again on the first of February, 1951. Two different ways of making membranes. The slight difference of shape in the former made a difference. So here I am, a calibrated thermometer was used to obtain the accurate density of tribromomethane, ChBr3, which was used for measuring pressure. So a piknometer was acid and alcohol cleaned and distill water filled with de-glassed distilled water and put in the thermostat at 25 degrees. And after two hours the level was adjusted to the lower mark, and it was weighed, and then it was emptied and dried, and then the tribromomethane was added in order to measure its density. So a mass with PYK, the piknometer, plus tribromomethane. Mass of aqueous contents, 27.5612 grams, and with the ChBr3, 79.5348 grams. And so the specific gravity was 2.886 corresponding to something here of 2.877. Unfortunately a yellow color, so there was some bromine formed. Have to wash again, dry thoroughly. I did it again and I got a density this time of 2886, which is the same as I had before. Making more lactoglobulin, surface tension. Review of osmotic pressures to date shows that .5, 1%, and 2% runs are really advisable in that order. Because it didn’t matter if they are more concentrated, it was diluted a little bit by the more dilute, but the dilute being… I felt that would make them more precise. Surface tension again, apparatus was cleaned with this nitric acid ethanol mixture and left in ethanol potassium hydroxide over the weekend, getting everything very clean. The blank was 0.022, which is five times better than I needed. And I’m satisfied that the blank is -0.05 plus or minus 0.07. Means that the error is greater than the blank itself. I don’t like that. There must be some error there. There’s a decimal point missing I think. Electrophoresis on June 11, 1951. That’s the end of that book, we’ve finished Book 4.