3 March, 1959

Dear Daddy,

Thank you for your letter of the 20th which I was pleased to receive.

Bookcases: wood is much cheaper in California than here, and I think rather nicer than steel if you can get someone to make them. I am quite satisfied with my steel shelves. They are cheaper than wood, at least any ready-made wood. I think I shall get some 9" library shelves from the same company, which are more expensive than my 12" light industrial shelving but worth it in the added available space in the room. The same company make a system of hanging lateral filing, looped pockets of heavy manila that hang under a special shelf in 12" widths for quarto and 15" for foolscap. I think I might get some of the quarto to hold reprints and such. I shall not get anything more until we move though. The 12" shelves work out about $2.00 per three foot shelf.

We have been looking for a place to live. I found a house at the corner of Ladbroke Road and Ladbroke Grove, (we have been there) on three floors with a 50 foot garden facing west, the owner has now withdrawn it from the market. I do not know what that means. It was too expensive anyway.

I shall probably write to you sometime about money. If we were to buy a flat on long lease or a house, and I think that we shall have to buy, I can borrow at least half the purchase price from the mortgage people at 6%, pay back over 20 years. Then I must find the rest and something to fix the place up somewhere else. I do not want to borrow more than £3000, this being about the maximum my salary will pay off. The interest I pay, by the way, is a tax deductable expense which saves 1/3. The rates may go down. I do not think we can find anything less than £4000 or £4500 and perhaps £500 to fix it up, and of course we do find places that are more.

I am sorry that my birthday letter to you talked all the time about presents for me rather than presents for you. Perhaps sometime I can write you a letter and not discuss me.

I designed some shelves for Linus seven years ago that used rather thick wooden shelves and then galvanized iron water pipe with holes drilled through it to support the shelves. It of course was a nuisance to drill the holes, but cheaper than brass. I think standard pipe flange fittings were put into it too. You may feel it is too dangerous.

I hoped we might have a new place for you to come see us, and a new bed, but it is unlikely.

I am doing a lot of teaching. We have the third year boys in the lab and I have four experiments. I must give my five lectures next term. What was that electron deficient compound you were telling me about?

I have a new 1/16 inch lead wall in my room to protect the new Gregorio's. I feel better. Actually I have a rather bad cold and try to rest at home. I went home early yesterday and went to bed, but have been house hunting this morning.

I might suggest to you that when you have available the details of your visit to England, that I write to the Home Office that you are coming and what you will be doing and requesting they make a decision and instruct their officials accordingly. You may think it a loss of principle, but it is the way things are done here and it would prevent having to hang around the airport. I would write to them as a resident.

Much love from,

Peter

3 March 1959

To: Linus Pauling

From: Linus Pauling

Subject: Von Neumann

John von Neumann in his book The Computer and the Brain discusses the natural nervous system. He says that the natural computer element, the neurone, has a rather slow reaction time, one one-hundredth of a second - that is, slow compared with modern artificial devices. He says or implies that the method of transmitting information by digital pulse trains, interpreted in terms of pulse-time density to give digital-type logical information but analog-type quantitative information, will reduce the effective computing speed by a factor of 10 for each effective decimal place of accuracy. On the other hand, he suggests that the effective memory capacity of the brain must be of the order of 10²⁰ decimal digits, in order to record a lifetime of experienced sensations. Hence the brain, compared with modern machines, is a very reliable and compact computer, with a low speed and a low order of accuracy in individual calculations. It has an enormous (inactive) memory capacity, and a large (active) arithmetical-logical assembly (highly parallel operation), and great economy in power requirements.

Von Neumann's argument is that since the brain has to control such a highly complex organization as the human body, it must carry out long calculations. The brain cannot work to 10 or 12 decimal accuracy, which we know to be necessary in artificial computers to retain two or three figure final accuracy, and hence he argues that some different and unknown form of mathematics must be employed in the brain to produce its satisfactory results. He says that the loss of accuracy at each stage of a long calculation may possibly be eliminated by carrying out a large number of independent calculations at each stage and evaluating the results statistically. Because of its large active assembly the brain is equipped to do this. The second point, suggested by the reviewer in Nature (A. H. Armstrong, Nature 183, 350 (1959)), is that the brain never does any arithmetically deep calculations (that is, long calculations), but merely makes minor interpolations, variations, and extensions to a vast stock of "tabulated solutions," which it has been accumulating on a basis of trial and error and statistical selection, in its memory throughout the course of its development from a single cell. Really basic problems, such as digestion of food and circulation of the blood, have perhaps been tabulated prenatally for a reference in this way. (I would point out the ability of the chick to interpret impressions from its eyes.) Slightly less basic problems, such as learning to walk, have not been solved prior to birth, and it takes the infant's brain-computer many months to solve this problem. Even the adult's brain-computer may take a week to solve the rather simple problem of riding a bicycle. When obtained, this grid of experimental results is stored in the memory for future use by simple interpolation.

Much of the above material is quoted verbatim from the review in Nature.