September 20, 1962

Dr. Luca Cavalli-§ forza
18, Via Fatebenesorel le
Milan, Italy

Dear Luca:

Would you care to computize a key-word index to

these SI8s? Put a man In loop to mark key words?
Later we may discuss better grouping techniques.

Sincerely,

Encl: SIB Nos. 3& 4
Notes on Preliminary test on Burroughs 220

a

ppt
vore tc Luca
idea on random nunber generator.

Count intervals >etween radioactive disintegrations. These are distributed around
am a poorly estinated expectation. 3ut compare each interval counted with the averaze
value of all previous intervals in the set. These should accurately fora random choices
of >=1, <= 0 for construction of sets of random bits. (These do not, of course, nave
to be generated in real tine while the computer is handlin: the vain prosran.)

what if t J, &, but = t,,7 I suppose reject that number.

Is electronic noise random enough to use this directly as source of pulses
without bothering about actual radio counter?

The idea of the average is to use the actual neasured distribution of past events
to set the threshold at best oureent estinate for p= 1/2. This say introduce sone
systematic errors in the early generation of the average, out theee snould soothe
out as the nunber of terms used to compute the average sounts. In fact one should

set the threshold not at the mean but the median time interval. (?a¢2 (che cedian is

also an average, so sy discussion above is not technically incorrect)

Use lons-lived isotope and so ignore gradual decay of counts,

How is echemo-, fluoro-, thera- poiesis?

| / —
P.G. Can you print your own compugerized library scan on a sheet seet up for sark-
sesste:? ihe a reader could make a check list from the full output index, ari feed

he 2 7 + 1 be . 3
this directly back to the compu Ler co prepare a selective oibliograophy. (And by a few
sucecessive approxiuatioss and a little lozic .o eatch duplications even prepare a
subject-organized index). This will »e fun to talk about in “oveaber; aeanwhile,
perhaps worthwhile to investisate tne vossibilities of a mark-sensins attaechuent to
reac the am edited output sheets,

P.3. what lansuace does your computer understand? Jo you have a coupiler? (1 have

yeen learning BALGOL).