March 24, 1950.

Mr. Gordon Allen,
155 Corona Avenue,
Pelham 65, N.Y.

Dear Gordon:

Thanks very much Zor writing me about your new experiments at such
a prelimigary stage. They sound very exciting, and I hope you can continue
to let me know of the results.

The method should be quite sound, and if you can get a clearcut selection
with the markers you hae, there ought to be no doubts in the results, with
the 5 provisdés you tabulated. I have a few suggestions of which you have
probably already thought;

Since s* is "Linked" to M, and Az™ to T, I wonder if you don't have a
more con’ ent way to pick principals and complementaries. That 4s, the
cross BMS x TL Ag®™ will give prototrophs which are mostly S9 As®. The com
plements will be S° Az’, which you should be able to select with relatively
little trouble. This would have the advantage that the principaas could, for
the most part, be picked out directly, rather than having to make further
testa on quasi-prototrophs to see which of them are P-B_-; rather, I should
say that the garkera are so arranged that a large propoftion of the prototrophs
will be 3” ag®, You undoubtedly have already developed such stocks, but
on the offchance you haven't, and that it would save you some time, I'm sending
W-1234 (W-677 Az™). I don't have a 58-161 5", unless Norton Zinder has saved
some, but you probably will want to use your BeMEP- anyhow. [P.3. He has: #1302]

As one of the first logical (not necessabily chronelogical) steps in this
type of analysis, I wonder af it would not be most impoartant to establish
the following types of stat&éstical complementarity among the population of
recombinants at large, namely that By * selections will show a distribution

of unselected markera which will be essentially complementary to the Pi.
selections, It may be better to use nutritional markers for this purpose,
although one could compare Az° S® prototrophs with Az™ s* “non-prototrophs”.
This may be important to do at an early stage, especially if the Mal locus
shows such an aberrant behavior, as I think it does, among prototrophs as
xkerorkkk among persistent diploids. The expectation might be that, in the
crosses above, Mal- might predominate both among the principals and the com-
plementariles.

W-677 has an excellent set of markers [In fact that is its raison-d'etre],
which may be more convenient to use than just nutritiona],. in SPO ZORY
3

eas
Forgotten: share araee aha Ree tent) pe ee end Thely more difficult.
If you need still more markers, you can use W-1272 which differs from “/~677 only
in carrying a Stl- (sorbitol) and a V6P (partially resistant tonT6; easily scored)
However, I don't have a W-1272 ST, and in fact, if you do develop one to use, would
aporec&até having one. I'm sending W~1272 just on the off chance as above.

Of the criteria you mentioned in your letter, No. 3 (Chrrelation of crossing-over)
should be the most feasible to test, especially if you use the variety of markers
availabls. Your eriterion 5, that "two complemantaries should not share a trait of
che principsl" is based on a rather more restrictive hypothesis cf the meiotic
mechanism than the others, but at any rate, will be rather hard to test exhaustively.
(You will always find a few "spurious" complenentarics.)

The picture is brightening just a b&it, not much, in my ower work. As you clearly
saw, I an convineed that Mal duplex prototrophs are the result of segregation from
a cell pure for Lac, etc., and heterozygous for Mal. While this night mean a "two-step"
reduction, by analogy with the persistent mexkaiemshexx "diploids" which are 2n for
Lac but 1n for Mal, another mechanism is also possible now, H-226 is a diploid
which is (exceptionclly) heterozygous for Lac and for Mal, [obtained not with Het
sgocks but by Lac,- x Lac4-.] Usually Lt segregates to give LaqrMal- and Lac ,—Mal+

Rather infrequently, H-226 prodtices apparant partial segregants, i.e., Lac v Mal-
for Mal v Lac-].However, unlike the Lac v Mal- which one usually obtains as the
typical persistent diploids, these "partial segregants" are homozygous for Mal, as
Mal+ reversions amon® them then segregate. H-226 is already homozygous for some
other factors, so that it is also at least one “partial segregation" removed from
the original zygote. The impression that I get is that some sort of autogamy does
oceur, almost invariably after the originak fusion, and occasionally thereafter,
which may result in the loss of heterozygosity for various factors, befdre segregation
to recognizable haploids occurs. I don't quite know how to fit this in with the
other phenomenon of hemizygosity (viz. of dal) which seems to occur quite often,
but this might be an accident of non-disjunction. I am about to test,now, to see
whether the descendants of H-226, in which the elimination for once did not occur,
may continue to show this more orthodox behavior in subsequent generations of

crosaing.

I don't have time to put down the details, but you can tell Bernie that I have
some evidence shortx from irradiating diploids, that UV-killing is partly or even
largely nuclear, but not, for the most part, recessive lethal mutations. However,
even after large doses of UV, a substantial fraction of the survivors can stil
segregate. In a sense, we are both right in our contentions as to the haploidizing
effects of UV.

Sincerely,

Joshua Lederberg