Avgeost 20, 1949.

ae. N. OD. Zinder,
Carnecte Institution,
Cold Soring Harbor, N.Y.

Dear Norton:

Iam sorry not to have been able to answer your letter soonsr, but have
just returned from a trip.

Your data certainly appear to he dnconzistent with ntre for the frequency
of the V,* Lae~ recoubination class, and I dontt see offhand just why they
should be. I would suggest that you yourself make a direct comparison of crosses
not involving S& with those which ac,

Ioen not clear how you crossed ths resistants with the dspencenisatIf the
cross wis carried ont on strentonyetn-lacting nedius, I tae it that you would
have the cppertunity cf recovering streptsycinesensitive recombiaanta unless sr
and sd were alielic, tnt ould Lose any ad prototrcpns. Tha lous of these ad
protetrophs wight be expected te bLlas the segregatioas of enay factors linked

to sd, which does seen to be happening (919 x sdl: you wrote this as Leet x Lact,
but I assume that you meant sal to be TLB;- Lac~ ¥)". In this instamce, the

near absence of V)* prototrcphe indiestes a linkage of sd te Ye) >

Tour suggestion that. er may have a "slight" recutreceut for streptomycin fits

the data given very well, but you should control your observations on the Las Vy
segregations with experiments on the stroptorgein sensitive paventes. It would be
dangerous to assum that the modified responses to Sleep iutlusasulie resulted from
maltiple allelism, without experizents crossing ad's ard crts with sss (wild type)
and looking for exchanges of the modified character (a.g. st x + night give some
recombinant sd‘'a, or the comvarse.) Is thts what you zesn by your cutcrosses?
With regard to crosses of sr x sd on streptomycin medium, I wonder whether you
have checked the sd parenta to ascertain whether they have any nutritional require-
ments beyond streptonyoin? £.Q., will adel grow on ainimal + TLB, + streptomycin?
Will sd x ad give prototrophs on steeptomycin medium? But alec remember that
linkage of sr to appropriate nutritional markers will also give all resistant
prototrophs. Instead of using heterozygotes, which are more complex than ever,
why not simply do reverse crosses (i.e. BM sr x TLB, + and compare with BM + x
BLB, sr)? I would be interested to try an experiment here with sr in a heterozygote,
primarily to ascertain doninance. Demerec has assented to this, and I would appre-
ciate it if you could ask hin to take along cne or two strains, It would not be
especially casy to determine cytoplasale inheritance with the heterosygotes at the
obey pong because of the possibility of hebhsy, gosity which would also prevent
segregation.
The more I have thought on the natter, the more convinced I am that we
should dispense with all extraneous iesues on the Salmonella program, and
devote a lot of time when you return to the production of mutants 4n one
Salmonella strain after another, until one is found with a workable recombina-
tion system. Therefore, I think it very well-considered that you are letting
the phage problen lie, and hope that we will not be terpted to resurrect
it. Don has been turning out mtants from new colf straine at a great rate,
and I think that the same procedure should be used with Salmonella.

L. Cavalli has written that he has discovered another coli strain which
recombines with X-12. The ice is being broken.

Cavalli has also discovered asderivative of 58161 which recombines very,
very frequantly, and using his "fr" stock, I have been able to do crosses
entirely on complete medium, and get data on unbiased segregations from
non-persistent zygotes. The results tie in very well with the conclusions from
the mapping of Mal and Gal in prototrophs, and their segregations &n the
persistent heterozygotes that the segregations are not normal. There isa very
definite elimination mechanism which limits the kinds of segregants and recom
binants which can be found. Complementary recombinations are not produced with
equel frequencies (if at all), and it is impossible now to accept a simple
interpretation of any mapping data. Therefore I would not take too seriously
your difficulties with Lac and V,, but agree with Dew rec that the important
point is the use of redombination to detect allelism. The situation may be
parallel to Auerbach's unstable centromere (Genetics Jan. '47) but Le still
very murky.

Sincerely,

Joshua Lederberg