Nevember 23, 1950.

De. Margaret Lieb,

Rerckhoff Laboratories,

California Institute of Technolory,
Pasadena 4, California.

Dear Peg:

Thank you for sending your thesis, which I return herewith. Have you
sent it to Genetica?

Your letter invited caments, so here goes: (from particulars to genoralities)

Unclear pointe: pe 13- whieh work of R & 3 (unpubl.)

Suumary: “maximum of 200x7 200x what?

Literature cited: Lederberg 194)b (heterozygotes) miasing (p.13)
Ryan 1950 4a en improper reference, ss G3 is not a
publication. (Asi Nelson about just this point). The
method haa been published before.

What is your policy concerning initials or names in

your oitations? You should be consistent. A generally
good policy is to use initials only for men, and the

full first name for women, except where there are many
authors . Zentr. Bakt. does not have be be spelled out
at such lengths, but the Abte (A or 3B; and Originale or
Refera$e should be indicated ( abbr.)

Arguable pointe: Newoombe (1943)ae00 ia rather inane, since the "formula"
doce not detect umphtk anything. You might write: the recovery of
mutants may be expressed as the ratio re/n» } Fon) )

,or Js

Fy/ny ryng

At best, the quoted arrangement is inverted for the use you want
to make of 4¢ in table 3.

 

 

"the entrance of histidine inko the zene istelf’... Tsk) (Shades
of “pantothenate impinsing on the cene"!) Of gourse you might
well and properly mean that protein building blocks, ineluding
histidine, arecnecessary fpr the building up of hf function. But
this sudden lamarckian note (4.e. direct relationships between
gene sonstitution and biological effect) is a little surprising.

Geheral discuseion: 1) Phenotypic lag in spontaneous rutation. Newcombe's
argument is on the whele very weak (although his conclusions are acceptable)
becsuse he presupposes a specific model of the mutation process. The discrepancies
between Methods I as against II and III hinge on the average clone size, dete,
ratio of mutant celle to mutant clones with one or more effective mutants. This
ratio ("d" in the atteched) can, of course beinfluenced by phenotypic lay, but
P33

it also depends on the model you use (as you have pointed out for the possibility
of nuclear secsregation). The Luria-Delbruck model supposes that each cell has
a smooth probability of mutation between each fission, at a rate proportional
to the fission rate.The individual cell ie assumed tt increase smoothly, rather
than discretely, from one to £2 two, at an exponential rate. A mutation occurring
early in an interfission interval thus has a yield of tus 2 mtetite at the end
of the interval, while one which oceurs late has a yield of &, the average yield
being .5. + 164 . Another (and perhaps simpler)model would be that mutation occurrec.
only at fission, with a yield of 1.0 (dees, one nommutant; one mutent progeny -
theory of mutation as copying error ) or of 2.0.These differences will be reflected
in a comparable change in the theoretical clone eize. (I forgot to mention
above that you ovsht to clarify, owlt, or otherwise modify p.8 line 6-8). Suxxten
Anothcr, less critical consideration, is that it 4s invaldd to strike an arithmetic
mean of method TI - but anparently you did not do this. The discrepancies of
Newcombe's data are exaggerated almost 2-fold by this error (whigh arises from
the assumptions of the "likely average" approach of LUD - one has to pool
all the data somehow, and use an overall "6" for the series). I think you would
have done better to use Lea and Coulson's maximum likelihood method in view of
the oritical use you wish to make of your data. While you are justified in
concluding that no diseresancy (which might be based on phenotypic lag) could
be detected, I don't think that you can say with even mild assurance that there
is no phenotynic lag fok the enentencous mutations, whch your summary at
least might misimply, (as well as the staterent " only induced hf mutations
exhibit phenotyric lag "). You probably do not intend to cenvey this impresaion,
but I sugrest that yeu look out for it yery carefully.
Enelosed is 4 mimeogrephed surmery, modified from LD which I've used in
classes. It shows the discrepant result of the model of mutations at fission
only. The diserepancies between I and III can be most objectively phrased in d.
Another pasring point: how wera the date of your tables weichted 7 (By numher
of observations, or by 1/sD¢ #9 or 1/sd4 ).

A point of considerable impogrtance concerns the dominance of hf. I don't
think you are ontitled to moke a very strict in*erence from the behavior of K-12
heterozygotes. You don't really know whether the mutation from hf to h- de a
"loss" or asain". But in any event, if you define h/ as dominant, it ie in-
consistent then to refer to phenotypic devadopment being obscured by the time
required for sesregation. Tho h- phenotyoe could only begin to develop after
segregation was completed, 4f h/ ia dominant in your sonse.

Along the same lines, I am not clear how far you are trying to generalize
your results on the relative rates of mutation from and to hé. Your discussion
is cagey, but your intreduction leads one to look for a feneral comparison.

It might be worthwhile to emphasize that (p.29) the spentancous mutation rates
of differont atocks from h- to hf straddle the rather consistent rates from

Your final conclusion (that I can discuss), that induced h- have no lag
is someifaat disturbing. Your suggestion 7.31 that the h? may be effectively
h- to begin with is nrobably correct, since the cells wera vrepared so aa to
be in lag phase. One must also consider that both UV and penicillin may actually
accelerate the dex loss of hf phenotype. Would it be possible to determine
whether uv treated hy are nore reaistant tp veniclilin initially than untreated,
or does the problem of liquid sensitives rule out a detergmination?

On the whole, the probles and troatment are very interesting, but it is un~
fortunate that the complexity of the material makes a conclusive determina tion
almost beyond reach, for most of the really interesting questions. That makes
critique easy, but proofs difficult.