Your letter just arrived forwarded from Washington, and, having recently a lot of practice in decoding proteins, I have decoded
your handwriting almost completely. I drove to Pasadena a fortnight ago, and spent two days with Jim, Max Delbrick [sic] an
others. They have a model of RNA, big and nice looking, but they do not believe in it very much themselves. (except of Alex
Rich who conceived it) It has trapezoidal holes [diagram] formed by two bases, and two different "sugar edges". And
there are 20 different holes. But I have found, however, that the combination rules do not work at all. According to that
model, 10 amino acids can accur [sic] only at even places in the protein sequence with the other 10 only in odd places, which
is certainly not true. In particular, dubbling [sic], like GluGlu, or CyCy, is not permited [sic] at all!
After I came back, I have tried a new code of Triangles (with three independent bases
[END PAGE ONE]
[BEGIN PAGE TWO]
in each corner) with the eye on the monostranded RNA (whatever the model could be!). There are also 20 triangles, but the
rule of combinations are different: 4 of them can combine with 10 others; 12 with 7 others, and 4 with only 5 others. I didn't
yet use this for insulin decoding. Will do soon.
But what I did was to take the insulin and vassopresiv [sic] data and to make statistics: how often (m) each amino acid accurs
[sic], and how many (n) different neighbours [sic] it has. Thus, for example, Cy accurs 7 time and has 10 different neighbour
[sic]. Plotting n against m one gets something like that [diagram] Amusing, isn't it!? Why slope 3/4!? I have tried to
derive theoretical formula, but couldn't, and the people from statistical Lab. here told me it is a full time work for
at least a month. Thus I tried another way:
[END PAGE TWO]
[BEGIN PAGE THREE]
: Sellected 60 letters (out of twenty) and arranged at random using tables of random numbers, and tried to make it up, first
in triangles, than in diamonds. In triangles it does not look at all like a protein: udles [sic] and udles [sic] of dublets
[sic], and triplets! In diamonds it looks much better, but I did not finish it yest [sic].
Jim comes here on Thursday, and will stay for several days. So I hope that he will join me, Calvin [footnote: chemist], and
McMillan [footnote: nucl. phys.] in building the Fisher-model of DNA since my strained rings cast-in-plastic just arrived.
Well, as you see from all said above, my opinion about all this "decoding stuff" is: promissing [sic], but not yet
quite right. Ceterum censeo DNA-PROTEIN esse delendam!
Would love to have a joint article with you and Jim on that subject.
Yours as ever: Geo.
P.S. Incidentally, due to a spontanious [sic] mutation, my longer article will appear not in the Nat. Acad. of Sc. of U.S.A
but in Royal Danish Acad. of Sc.
Please write air mail to the above address