Jane 11

Dear Josh,

I agree with your comments for the most part. We had predicted that
the population would "snap" from one predominant component to another
and that the average ratio over a very long time would be that
expected from mutation @nd specific selection), as in your graph.

If each adaptive leap is due to one mutation we would expect a
changeover, under the conditions of our experiments, once in 40,000
years on the average. However, our experiments suggest that actually
a large number of mutations are involved , since the variance of

cycle length is quite small and there appear to be appreciable numbers
of the, advantageoud mutang present when selection first begins. If
say 10° mutants must accumulate to initiate selection, then in our
case some of these would originate from h+ about once in 4 years.

But when this occurred it would not result in a changeover, but rather
in a small temporary increase in the equilibrium ratio. Cf course

any sort of effect cculd result from a sequence of improbable events,
& point which is often made in ponular articles on statistical mechanics
but which we can ignore for practical purposes. A curious feature

of periodic selection in our case is that the immediate cause of the
adaptive leaps operates in such a way that the increase and decrease
of minority mutants is approximately complementary, so that a fairly
flat equilibrium 1s simulated at all times. We know this not only
because an equilibrium is in fact observed, but &lso because we can
observe the increase in ht starting from a small. h- inoculum and in
other experiments the decrease of marked ht from equilibrium and we
find that these independently observed increases and decreases are

in fact approximately (sometimes exactly) complementary. In this
respect our periodic selection is subtly different from other
non-specific killing agents. Thus both in the (reasonably) long

run and in the short run a condition of equilibrium can be brought
about by periodic selection.

If you spend a few more hours thought I am convinced that you will
be convinced that your statement that periodic selection cannot
possibly influence the probability that a single X gene will be

+ or - is erroneous for practical purposes, although correct if
we sum these probabilities over infinite time. The same holds for
your statement that periodic selection will simply broaden the
dispersion of the distribution of mutants.

I am planning a theoretical paper embodying your Suggestions, the
above considerations , examples in the literature etc. for the

Am. Nat. soon, and I will send you this one before it is submitted.
Your suspicions are right about my weakness of character. I won't
recall the present one. Its just a preliminary note anyway, and in
my opinion no paper can be classed as superb whose content deals
with events solely on a populational level. We have changed the
authorship to &@, S, & Ryan after convincing him that our use of his
data warranted this. At first I couldn't tell whether F.J. was
Just being magnanimous or doubted the validity of our interpretation.
It turned out to be the former.

Actually we know a lot more about periodic selection than is stated
=2-

in our ms. We now know that colicins are the cause of most, but

not all, of the adaptive leaps. Also a population under conditions
of stockkeeping rapidly becomes heterogeneous with respect to
resistance to its own adaptive potential. Thus if we start a ht/h-=
reconstruction with stocks "in phase", from 10 to 20% of the
indicator population drops out in the first few transfers and the
rest drop out at the next adaptive leap some 30 transfers later.
Some experiments we have going now indicate (I mean deronstrate)
that the many minor fluctuations in +/= ratio which occur between
gajor adaptive leaps are also due to non-specific factors inherent
in the system. When we start with h-m- and follow ht and mt in

the same STs we obtain two equilibria (m+ is a little higher than
h+) and the fluctuations in these are uniforrly parallel;

Cultures from survivors of radiation are often out of phase with

the parental culture, as we found out in trying some induced mutants
we got from Peg. One of her stable h- mutants is giving a very
peculiar result. It remains stable for a (reproducible) period of
20 transfers then produces ht which rise to a peak then completely
disappear within the next 10 transfers, I'll keep you posted on this
series. Finally, we suspect that each adaptive leap may be caused
by several different kinds of mutations in the major component which
are about equilly suitable as selectors against the antecedent
population without too much interference with each other, I'll give
reasons for this if we confirm it.

By the way, I have turned up several mutants in Neurospora which can
be most easily interpreted as inhibitor producers. They are in the
amycelial component of an ornithineless vs. methionineless-amycelial
heterokaryon. The behaviour is as follows: An isolate from UV
treated conidia gives conidia which form n essentially normal colonies
on minimal, but no normal colonies and more than n amycelials on
methionine medium. In two cases these are associated with lethala

So that we get normal colonies on minimal but nothing at all on
methionine: Size comparisons show that the normal colonies in all
cases grow unusually slowly.

In UV effect in Neurosp. I trust the kinetics, which show conclusively
that the effect is entirely nuclear except possibly at survival of
below .01 . Norman has completed a lot rore work on this since we
discovered a way to vary the mean number of nuclei at will (by altering
the medium). The extrapolates for conidia with 203, 4.5, and 5.6
respectively are precisely these numbers except that conidia with

2.5 nuclei show an extrapolate of 2.1. This was shown by Norman to
be due to the uninucleate component they contain which has about

twice the inactivation cross-section ber nucleus as multinucleate
cells. All curves for microconidia &o right through one under the
conditions of his experiments. He has ruled out shielding as the
cause of the increased inactivation pepss-section per nucleus in
macroconidia. I have shown in two different ways that the probability
of homology of recessive lethals is negligible. Therefore the

killing of macroconidia must be due to some other effect, and the
inactivation cross-section of microconidia (per nucleus) must be
larger because recessive lethals are expressed immediately. Last
spring I was worried about this interpretation because the kinetics
-3-

of recessive lethal production by the heterokaryon method were
not first order. I have stopped worrying about this now because
in a large number of later experiments the kinetics vary from
perfect first order to pronounced maxima with induction pertods
and declines depending upon uncontrolled conditions of the
experiments. There is a beautiful trend throughout. The greater
the proportion of mutant nuclei recovered (I have up to 80% now!)
the more nearly first order kinetics are observed. A clue to
what was happening turned up in an experiment in which I counted
mutant nuclei in both components of the same set of isolates.

The frequency of isolates with lethals in both components was
about three times that of isolates with lethals in one component.
Barring unforeseen inhomogeneities in the population, this can
only mean that the observed mutants are the residuum of recovery
from @ more general damage to the genetic material. This initial
damage has first order kinetics, and this relates the findings
above satisfactorily. Experiments with which you are already
familiar show that the non-genetic effect is reparable in the
presence of an unagéfected nucleus. The problem of UV inactivation
of Ne. conidia seems to be considerably clarified, but I would
appreciate any queations you may have. I will have two papers on this
finished within about a month and will send them to you.

Turning to your case, I would place a gocd deal more confidence in
the inactivation kinetics than you do, despite the Lwoff effect.

As I recall they are the same for the heterozygote as for ordinary
Klé, and are high multi-hit affairs. This, along with ycur failure
to find balanced leéhnals strikes me as significant. I would suess
that the latter is due to duplicity of each comvonent of the
heterozycote. After all, lethal mutations must be more frequent
than screenable ones, yet you are not picking therm up. A multiplication
of genetic sets might result in sufficient resistance of the
bacteria to this kind of killing so that cytoplasmic killing would
start at relatively high survival and lead to a very high hit number.
A kinetic analysis of induced isogenicity might be helpful.

The army is really closing in on me now. Apparently Columbia U. lacks
any influence with the medical review board despite many letters

of protest including one from the Committee on Govt. aided Researech
explaining that I am now principle investigator on two grants. I
learned yesterday that I am still in 1A after review. Of course I
wouldn't like to work in BW, but would do so to protect our boys in
service from my blundering attempts at therapy. Do you think I should
write to Krueger & Braun? Or is there anything else I can do?

Sorry to hear Bussard is homesick, but its natural under the
circumstances. I'd be too. KCA IV is very far advanced for 43
months. I hope we'll see you at CSH this spring if not before.

Best to you and Esther,

o ~ —. . :
‘ ™~ ‘ — r y o v ee on . é. tds / q ie . i ra é vo eee ee . - é fy