Skip to main content

Full text of "Proceedings"

See other formats


twenty-five years of progress 
in mammalian genetics 

and cancer 

Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 












SOTR^p, R00]vf 


Proceedings : 
Symposium on 
25 Years of Progress in 
Mammalian Genetics 
and Cancer 


Koscoe B. Jackson Memorial Laboratory 
Bar Harbor, Maine 
June 27 to 30, 1954 

Edited by 
Elizabeth ShiilT Russell 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 
316263—54 17 

Introductory Remarks 

Dr. E. B. Wilson, President, Board of Scientific Directors 
Office of Naval Research, Boston, Mass. 

I have the duty, the privilege, and the pleasure of welcoming you to this symposium; 
I shall be brief. Let me say first that I am glad to be here myself and glad to see so 
many of you in attendance. 

Twenty-five years is a short time in the history of old institutions like Harvard or 
the University of Cambridge or that of Bologna. We are young, but I venture the 
opinion that in our 25 years we have discovered new truth in excess of that discovered 
in any of those three institutions in their first 25 years. And our future looks bright. 

In the early days of this century when the subject of physical chemistry was new, 
it was commonly said that the physicist, though making most precise measurements, 
was indifferent to the constitution of the materials measured, and the chemist, while 
exercising great pains to obtain his materials in their purest form, was indifferent to 
precise measurements upon them; but the physical chemist was indifferent alike to 
the purity of his materials and to the accuracy of his measurements upon them. That 
period is now so far past that it is often the physical chemist who has the purest mate- 
rials and makes the most precise determinations upon them. 

Such pejorative comments are not unfamiliar in the history of science in reference 
to new and interdisciplinary fields of research. At the beginning of this century they 
could be heard about genetics from old-school biologists of various sorts when the sub- 
ject was not yet accepted or even widely known. But in especial reference to the topic 
for this morning's session: "Inbred Strains as Research Tools," I may point out that 
one of the great accomplishments of the past quarter century by this Laboratory is 
its contribution to a general recognition that we must have pure strains of animals 
with which to experiment, as well as good techniques of experimentation, if we are to 
get satsifactorily definite biological results. 

With these few words, I will stop these introductory remarks and let this symposium 
get down to the business for which we are here. 


Grateful acknowledgment is made of the generosity of the Rockefeller Foundation 
in making this meeting possible. 

These Proceedings received for publication September 10, 1954 



Session I. Inbred Strains as Research 

Chairman, Dr. William S. Murray, Administra- 
tive Director, Roscoe B. Jackson Memorial Labora- 
tory, Bar Harbor, Maine 

(This session is dedicated to Dr. Clarence C. 
Little and Dr. Elizabeth Fekete) 

Speaker: Dr. John W. Gowen 

Significance and Utilization of Animal Individuality in Disease Research, 

Honoring Dr. Clarence C. Little 
Discusser: Dr. George Jay 

Speaker: Dr. Thelma B. Dunn 

The Importance oj Differences in Morphology in Inbred Strains: 

Honoring Dr. Elizabeth Fekete 
Discusser: Dr. Edwin Murphy 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 
316263—54 10 

Significance and Utilization of Animal 
Individuality in Disease Research *• 2 

John W. Gowen, Iowa State College, Ames, Iowa 

As the first speaker on this program, and as one who saw darkly the 
germination of the ideas that led to this notable laboratory, I may be 
pardoned if I recall some events which resulted in its founding. The 
program designates this symposium as the celebration of the 25th anniver- 
sary. But to me, this laboratory was conceived and initial steps were 
taken for its inception several years earlier when University of Maine 
students used to gather in what was then the pine-surrounded frame 
buildings under the lee of Pickett Mountain. Here, under the stimulus of 
Dr. Little's personality, these students became imbued with an interest in 

It is little wonder that, leader that he is, this beginning should have 
become the cradle for, and the inception of, a leading institution in cancer 
research. The wonder is not in the inception of the laboratory but in its 
accomplishment. A well established laboratory equipped with all modern 
features of research and with a staff of excellent people is relatively easy 

J to maintain with able leadership. But Dr. Little had no laboratory or 

visible means for one. We are celebrating the bringing forth of this 
laboratory literally from the rocks of Mount Desert. How difficult it was 
to start from the bare ground and build bit by bit this magnificent 
organization, only Dr. Little and the group of scientists who built with him 
really know. Each step was fraught with financial impediments beyond 
easy conception and to add to it all, the laboratory had actually to rise 
from its own ashes. 

The research emphasis of the staff is close to the theme I am to discuss — 
the individuality of the animal and its significance to medical research. 
The laboratory pioneered the utilization of this individuality by the forma- 
tion and distribution of their inbred lines to the medical research world. 
Their efforts made the medical investigator individuality conscious. All 
science is indebted to the small group who gathered here under Dr. Little's 
leadership a quarter of a century ago. Despite financial depression and 
fire, they contributed much to science and to the opportunities for 
research we now visualize in this unique laboratory. 

The research I shall cite deals with the effects of this individuality of host 
and pathogen in infectious disease but the principles developed have broad 

* Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine; 
June 27, 1954. 
2 Journal Paper No. J 2570 of the Iowa Agricultural Experiment Station, Ames, Iowa. Project No. 1187. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 


556 proceedings: symposium on 25 years of 

applications. The work represents the joint efforts of research workers 
whose investigations have centered at Iowa State College. 

A wild population, i.e. mice or poultry, resembles the population of 
human beings in consisting of a conglomerate of many different inheritance 
types. The variations ultimately depend on gene differences which are 
expressed through differences in the effects of allelic series within loci 
and/or in the interactions of genes occupying different loci. These varia- 
tions among animals within an experimental population have been a main 
cause for erroneous interpretations in disease research. Yet if this indi- 
viduality be controlled within groups while allowed full expression between 
groups it offers a useful means for the separation and analysis of the factors 
pertinent to a given disease. Several methods for forming such popula- 
tions have been devised — selection with inbreeding, close inbreeding, and 
formation of homozygous strains through outcrossing are representative 
techniques. On the genetic side homogeneity within groups has been 
emphasized over a period of 50 years of research. The early studies 
demonstrated differences in strain reactions to both spontaneous and 
implanted tumors in mice and rats, and to such infectious diseases as those 
of tuberculosis in guinea pigs and typhoid, Salmonella typhimurium in 
mice, or Salmonella gallinarum in the fowl. 

Selection of Species 

Selection, at the species level, has been practiced from the beginning of 
active experimental studies of disease. The species chosen must be 
receptive to the disease organism. The success of this practice is known 
to us all. In recent years it has taken another important step in the blind 
passage of pathogens through a series of animals within the species in the 
hope of adapting the pathogen to the species so that the adapted type may 
be utilized in studies of the pathology and immunity connected with the 
disease. This represents only another form of the utilization of indivi- 
duality in that the adaptations observed are in the pathogen and are 
presumably brought about by germinal or somatic segregation and/or 

Selection Within Species 

The selection of reproducible genotypes within a species utilized for 
experimentation is practiced by fewer investigators. Yet the variation 
between animals within a species seriously interferes with or invalidates 
the interpretation of many experiments. The host animals vary in genetic 
resistance to constitutional diseases, infectious diseases, and diseases of 
unknown antecedents. Where a pathogen is necessary the individuals 
comprising the population will vary in their virulences, immunizing 
abilities, and other significant properties. The differences observed are 
hereditary. The factors leading to the differences in heredity — changes 
in gene frequency in the contending populations — are mutation, selection, 
migration, and random gene drift. The factors to be considered in the 
distribution of the genes within the population are whether the individuals 

Journal of the National Cancer Institute 



are haploid, diploid, or polyploid, and whether sexual fusion occurs 
followed by gene reduction, crossing over, and segregation. Variation 
must be expected. In essence each animal is an individual apart from 
every other. The experiments must deal with the individual and yet have 
valid observations capable of application to the whole population. 

In disease research the pathogen itself may become the agent to separate 
out and differentiate between strains. A population of mice was exposed 
to intraperitoneal inoculation of S. typhimurium, 50,000 organisms [Schott 
(i, 2), Hetzer (8), Lambert (4), Go wen (5)]. Survivors were chosen for 
further breeding. This was repeated for six generations. The inoculation 
dose was then raised to 200,000 organisms and continued for eight more 
generations. In the original population only 18 percent survived. Even 
the black death of the 16th century was not that catastrophic. But why 
did 18 percent survive when exposure was uniform throughout the popula- 
tion? Fourteen generations later, the population was uniformly exposed 
to an even greater contagion. Instead of 18 percent surviving, 93 percent 
survived. Instead of survival being an expression of individuality sporad- 
ically appearing in the population, it has become the commonplace, and 
susceptibility is even more the characteristic of individuality. The 
changes in the resistances of the successive generations to S. typhimurium 
are shown in text-figure 1. That this change was no accident was shown 
by the fact that similar experiments with poultry [Lambert (6, 7) and 
Lambert and Knox (8-10)] led to like results. But there are other ways 
of utilizing the inheritance effects on disease. 







S 40 

u 20 


«*""— — . 





" ' 2 3 

<I0 5 ^ 


2 X I0 5 \ 


1 . 
1 / 

' 5X 

10 4 

1 / 

1 / 

6 8 10 12 


I 4 



Text-figure 1. — Changes in the survival of successive generations of mice infected 
with mouse typhoid, Salmonella typhimurium, as a consequence of selection for 

Inbreeding as a Means of Differentiating and Purifying Types Within 

a Population 

A system of consanguineous matings continued for a series of generations 
tends to make the individuals within a single family line more and more 
alike, the ultimate being the similarity observed in truly identical twins, 
as in the human being. The pattern of change is shown in the following 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 


Text-figure 2. — Direction of individuality into family lines through a system of 
consanguineous matings. Dots in group represent progeny range in expression of 

diagram, text-figure 2. The variation in reaction of the individuals 
within family groups, where random mating is practiced, may range from 
over the entire range of the character to only a narrow span depending 
on the family. There is no means of foreseeing what the family reaction 
will be. This condition is illustrated in the families of the first generation. 
With inbreeding and the later generations, the individuals within families 
tend to become more and more restricted in their variations within 
families. The individuals react alike within the family strain so their 
reactions can be predicted. For all the families which may be developed, 
however, the variation among families will be as great as ever. Nothing 
is lost in this study of the disease reaction when many families are used. 
These results are illustrated in the last generation of the consanguineous 
mating scheme in text-figure 2. 

Inbred lines of mice were formed by 20 or more generations of close 
consanguineous matings, brother X sister. They were then tested for 
their resistance to mouse typhoid, S. typhimurium. All grades in typhoid 
resistance were observed between them, but within each strain the re- 
sistance from test to test was fairly uniform. Selection between these 
strains established strains of known disease resistance to a standard dose 
of the disease organism as shown in table 1. 

The mice of table 1 were, before the inbreeding started, samples from 
the large population of laboratory mice probably having a survival rate 
to S. typhimurium like that of the mice shown earlier: 15 to 20 percent. 
The fact that is striking is that it was possible to select pairs of individuals 

Journal of the National Cancer Institute 



from this population whose 25+ generation progenies were differentiated 
so markedly in their natural resistance. The natural resistance varies 
from 88 percent in one strain to practically zero in another. Each strain 
has a characteristic level of resistance which it has now maintained over a 
period of some 40 generations without the parents of any generation 
having any contact with the disease. No selection for or against natural 
resistance to the disease was practiced at any time. The resistance of the 
strain is a characteristic of it, just as its eye color or coat color is charac- 
teristic of it, bred into it through the fixation of the inheritance brought 
about by the genetic breeding methods practiced in constructing each 

Table 1. — Natural resistance of strains of 
mice to BOO, 000 S. typhi murium 11C, intra- 
peritoneal inoculation* 


Total mice 























*Data obtained at the Genetics Laboratory, Iowa 
State College, over a period of 15 years. 

The sporadic exceptions of the original population of mice have been 
expaoded to where they are the overwhelming representative types. At 
one end a major disaster results from a few bacteria being introduced into 
the given mouse strain. At the other end the mice of that strain receive 
the pathogens with impunity. Between these extremes there are five 
other strains, each with a characteristic resistance but taken together, 
spanning the range from resistance to susceptibility. The fixation of 
these seven disease-resistant types shows that the hypothesis utilized 
was oversimplified. Many gene pairs were certainly involved, but each 
gene pair was not equivalent. Each had a special function which con- 
tributed to resistance or susceptibility to a greater or lesser extent, not 
uniformly equal extent. 

Inheritance in Disease Resistance 

Crosses of strain S with strains L and Ba, made by Hetzer (3) , demon- 
strated again the individuality of the strains and the inheritance of sus- 
ceptibility and resistance to this disease, table 2. 

The hybrids resemble the resistant stock in their reaction to the typhoid 
organism, with the F! slightly less resistant. The progenies of crosses of 
the susceptible strains L and Ba indicate that at least some of the sus- 
ceptibility genes in these two strains are affecting different characters for 
resistance. The results show the complexity of the inheritance patterns 

Vol. 15, No. 3, December 1954 

560 proceedings: SYMPOSIUM ON 25 YEARS of 

that result in individuality. Individuality in disease reaction is de- 
pendent on many gene pairs. Simple Mendelian ratios are to be expected 
from hybrids only in rare cases. 

Table 2. — Survival of progenies from reciprocal 

crosses of resistant strain S and susceptible strains 

L and Ba, 200,000 S. typhimurium 11C 
































The Cellular Dependence of Individuality in Natural Resistance to 

Mouse Typhoid 

Two hypotheses have been suggested to account for individuality in 
resistance to a disease. The hypothesis which has been followed is that 
inherited genetic factors in their chance combinations control the resistant 
individuals and account for their sporadic appearance. The other hy- 
pothesis is that the individual acquires an active or passive immunity or 
sensitivity. Some are supposed to retain a latent stimulus which is 
thought to keep up the resistance. These carriers may spread the sub- 
lethal stimulus to their progeny. Evidence indicates that passive transfer 
from mother to progeny may take place, but males are not able to transmit 
the passive sensitization to their offspring. The fact that reciprocal 
crosses of the susceptible and resistant strains react similarly (table 2) is 
against this scheme being an important method. Another method of 
testing this scheme for explaining the individuality of the reactions to 
disease (11) consists of crossing two strains of mice, the S and L strains of 
table 2, which have been differentiated in resistance and also carry coat 
colors by which the strains and their hybrids may be distinguished. The 
S animals have white, the L silver, and the hybrid black fur. One L 
female is mated to both S and L males in the same estrus. Such a female 
may have both L and hybrid progeny in the same litter. These progeny 
are subjected to all the common environmental influences from the 
mother and within the same uterus. If the acquired immunity scheme 
was correct the pure L silver young and the black hybrids should be 
equally susceptible. The figures show that they are not. The 
silver progeny were all susceptible while the hybrid black progeny gave 
the typical hybrid reaction of 47 percent resistant. Curves showing the 
progress after infection are equally distinct. The individuality that has 
been concentrated into these strains through genetic techniques represents 
rare inheritance patterns observed as sporadic types in the original 
populations. The question arises, what are the attributes that cause 
these strains to be so dissimilar in their susceptibility to this disease? 

Journal of the National Cancer Institute 


Biological Characteristics of Significance to Resistance and Suscepti- 

The problem of what makes one susceptible or resistant concerns us all. 
The data presented earlier show that the differences are dependent on 
many gene pairs. Of the characters differentiating individuals, some are 
commonplace, as weight and growth. 

The influence of weight and growth on resistance to S. typhimurium has 
been investigated for the strains of table 1 [Grahn {12-15)]. The weights 
and growths from 40 to 60 days of age were investigated. The mice were 
infected at 60 days of age. The strain showing the greatest survival is of 
intermediate weight. Below and above this strain are body weights of 
the most susceptible and the next most susceptible strains. The strain 
showing second highest resistance has the greatest weight, while the strain 
which is the third most susceptible is noticeably smallest. About 40 
percent of the total variation in body weight is due to genetic differences. 
Another 20 percent is attributable to sex influences. The weights and 
growth patterns prior to their infection are unique attributes of these 
strains but have little importance to natural resistance. 

Internal organ characteristics may be related to individuality of the 
different strains and to their disease resistances. Mice were examined for 
strain differences in heart, kidney, liver, spleen, and testis weights. 
Genetic differences account for some 40 percent of the total variation in 
the weights of these organs. For the kidneys another 40 percent was 
accounted for by sex differences. The livers, spleens, and hearts showed 
little sex effects. 

Hearts, kidneys, and livers of larger size favored survival. Spleen size 
was indifferent. The masses of the hearts and livers are more indicative 
of resistance than the relative sizes of these organs. For the kidney both 
the absolute and relative weights are significant in the prognosis of the 
disease outcome. The spleen-weight : body-weight ratio has a high 
negative correlation with the resistance of five strains but the correlation 
is broken by the behavior of the S mice. 

Resistance to body-weight change during infection is another index of the 
general well being. In S. typhimurium infection by 200,000 of 11C or- 
ganisms, the first 4 days following exposure are most critical. 

During the course of the disease both resistant and susceptible mice lost 
weight during the 4 days following infection [Thompson (16)]. Normal 
body weight for age was regained by the resistant mice generally by the 
7th day. These weight changes were correlated with the strains' natural 
resistances. The weight changes have real prognostic value but may be 
considered a result rather than a cause of the natural resistance. 

Change in strain spleen weights with the progress of the typhoid disease 
(16) show changes in spleen mass which are characteristic of each strain. 
The course of the disease was marked by enlargement of the spleen in all 
strains. The E strain showed least enlargement — 1.75 times the original 
size. The Z and K strains showed the greatest increase — 2.6 times. 
The mean spleen weight of the four resistant strains doubled by the 4th 

Vol. IS, No. 3, December 1954 

562 proceedings: symposium on 25 years of 

day and had quadrupled by the 14th day of infection, but these changes 
were not correlated with the differences in the strain resistances to typhoid. 

Humoral Elements in Individuality of Disease Resistance 

Circulating antibodies and the phagocytic cells of the blood have become 
major considerations to all problems of disease, however without much 
consideration of individuality in response. The blood volumes of mice 
vary over a rather wide range. When this variation is partitioned into 
that for the different strains, the variation between strains becomes rather 
large and that between mice within strains is much reduced. Plasma 
volume showed a tendency to rise, the rise being accompanied by a fall 
in albumin and serum proteins (16). 

The serum proteins of the different strains display unique characteristics. 
Of the seven strains analyzed only the E strain was observed to have 
beta-1 globulin (text-fig. 3). The concentration of this protein was 
sufficient to account for the increased serum-protein value found in this 
strain. During mouse typhoid, the beta-1 globulin increased (16) , patterns 
4 and 6. 

The inheritance of this beta-1 protein was studied in the cross of the 
E X S mice infected with typhoid. Sera were obtained from the four kinds 
of progeny: The males and females derived from the two reciprocal crosses. 
The beta-1 peak was present in all four sera indicating dominance of the 
gene or genes responsible for organizing this globulin. 

A new protein of a different type, alpha-l globulin, was observed only in 
the sera from S mice. 

Agglutinating antibodies to S. typhimurium were located in the gamma 
globulins (16). Electrophoretic analyses of the sera after absorption with 
S. typhimurium showed a gramma-globulin decrease from 20 percent to 5 
percent in the Z serum and from 8 percent to 3 percent in S serum. 

Cellular Elements in Individuality of Disease Resistance 

The leukocytes of the blood have been considered, since MetchnikofTs 
observation in 1892 of their phagocytic functions, a primary mechanism of 
disease resistance. In a population of mice there is a pronounced individ- 
uality in the number of leukocytes found in their blood. These differences 
were directed into strain differences through the genetic techniques used in 
forming the strains. Three sets of confirming observations have been 
made on these strains, two of which may be cited — one in the summer of 
1942 by Go wen and Calhoun (17) on young mice; the other by Thompson 
(16) in the winter of 1952 on older mice (text-fig. 4) . Both sets of observa- 
tions show that the leukocytes found in the blood of these strains are 
closely correlated to natural resistance. The leukocyte numbers in the 
two groups show a consistent difference, however, the numbers being 
greater in the 1942 observations. This difference could be due to age, to 
venous versus arterial blood, to season or to some unknown cause. 

There is a progressive change in the strain leukocyte number with re- 
sistance in the two sets of data. The two most resistant strains have the 

Journal of the National Cancer Institute 



-< Descending 

8 -i Q/tf or A 
Ascending ► 

Text-figure 3. — Electrophoretic patterns of 6 sera drawn from normal mice or mice 
infected with Salmonella typhimurium. Each sample contained the pooled sera of 
at least 10 female mice. 

Pattern 1 : Strain E — normal serum. 

Pattern 2: Strain E — serum drawn 2 days after inoculation of 2 X 10 5 typhoid 

Pattern 3: Strain E — serum drawn 3 days after inoculation. 

Pattern 4: Strain E — serum drawn 4 days after inoculation. 

Pattern 5 : Hybrid (S X E) serum drawn 4 days after inoculation. 

Pattern 6: Strain E — serum drawn 21 days after inoculation. 
The beta-1 peak was observed only in strain E and its hybrid progeny. In normal 
serum, the concentration of this globulin was low and its resolution was variable. 
However, during mouse typhoid, this beta-1 globulin increased to twice its control 
level. Electrophoretic analyses of 4 sera drawn from hybrid (E X S) mice at this 
time indicated that the gene or genes for this trait were dominant. From Thomp- 
son's figure 6. 

largest leukocyte numbers; Z and K strains less, and the E strain still less. 
The most susceptible strains, L and Ba, have fewest leukocytes of all. 

The strain differences in leukocytes extend into the changes accompany- 
ing the course of the disease (16). The leukocytes of the bloods of all 
strains show a reduction in number as the disease becomes more acute, 
the greatest change coming on the second day after infection. The strains 
with the larger leukocyte numbers retain their initial advantages through- 
out the course of the disease. As Thompson points out, it is difficult to 
reconcile decreases in lymphocytes with differential increases in the glob- 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 



















Text-figure 4. — Relationship between strain resistance and number of leukocytes in 
mouse blood. Solid line — Gowen and Calhoun, 1942 (17); dotted line — Thompson, 
1952 (16). 

ulins of the same bloods. Differential rates of recovery of leukocyte 
numbers in the bloods of the different strains became evident after the 
disease had progressed for 1 week. The S strain followed by the El re- 
covered first. The strains of intermediate resistance recovered more 
slowly. Counts of the different kinds of leukocytes were not important. 
It was as if the leukocytes came from one common stem-cell and that it was 
the capacity of these cells to divide to form large numbers of leukocytes 
when necessary that was important. 

The interpretations presented above are subject to further analysis by 
means of X-ray irradiation of the animal. These observations emphasize 
the careful consideration that should be given to any X-ray treatment. 
Gowen and Zelle (18) X-rayed 789 mice belonging to our 6 different strains 
with dosages ranging from to 700 roentgens incident to the body. All 
6 strains reacted in a comparable manner. Radiation decreased the 

Journal of the National Cancer Institute 


typhoid resistance of the various strains at the same rate per unit of dosage. 
The numbers of leukocytes of the different strains were decreased to the 
same relative amounts by comparable X-ray dosages. It is known that 
total-body irradiation will affect many types of body cells. It cannot, 
therefore, be concluded that the leukocytes are the only cells which have 
been affected and that they alone are responsible for the changes in re- 
sistance. Rather they are an index to what is taking place in cells of like 
X-ray susceptibility throughout the body. The X-ray treatments reduced 
the typhoid resistance of the animal and the numbers of its leukocytes 
proportionally over the full dosage range. While the leukocytes are not 
the only cells affected, these facts support the conclusion that natural 
disease resistance is dependent on cellular function and on numbers of the 
phagocytic cells. These relations furnish independent evidence for the 
conclusions reached by other methods. 

Parenthetically a large or small number of leukocytes does not surely 
lead to resistance or susceptibility to S. typhimurium. Weir (19, 20) has 
experiments in progress in which he selectively bred for high- and low- 
leukocyte numbers of the blood. After several generations the two selec- 
tions had diverged considerably in leukocyte numbers of the later genera- 
tion mice. Tests of these lines for resistance to S. typhimurium indicated 
that if they differed in resistance it was toward the lower leukocyte line 
being more resistant. The explanation of this result is not evident but 
could be a difference in the quality of the cells — as appears for the macro- 
phages cited in the next paragraph. 

Other fixed cells of the body, such as the macrophages of the liver and 
spleen, were shown to play a significant role by Oakberg (21). In suscept- 
ible mice, bacteria are readily ingested by the macrophages and large 
numbers of them may be observed within these cells. These bacteria 
appear normal, have good staining properties and appear to be reproducing 
normally. In genetically fully susceptible mice the bacteria seem to 
increase within the macrophages to the point where they may break out 
of the cell and become new foci of infection. In resistant mice it was 
difficult to demonstrate macrophages containing ingested bacteria. This 
was only possiWe when the dose given was 100 times that received by the 
susceptible mice. In these cases the bacteria do not stain well and their 
cell walls appear ragged . It appears as though the macrophages of the 
resistant lines have a highly effective digestive enzyme which rapidly 
destroys the ingested <S. typhimuriums, whereas the enzyme is in reduced 
amounts or absent in the macrophages of the genetically susceptible mice. 

The liver cells of the resistant strains will perform the vital functions 
in the glycogen cycle and in fat synthesis even in the presence of large 
lesions, whereas the liver cells of the susceptible mice will not. Resistant 
strains may show extensive lesions of the liver and survive, whereas the 
susceptible strains will die without any clinically evident damage to that 
organ. By contrast, the naturally resistant mice show but little damage 
to their spleens although the disease may be severe. Susceptible mice, on 
the other hand, ordinarily display noticeable lesions in this organ. These 

Vol. IS, No. 3, December 1954 

566 proceedings: symposium on 25 years of 

observations again call attention to the capacities residing within the cells 
which are of significance to the disease resistance. The liver cells of re- 
sistant mice are able to wall off the large necrotic lesion, block off the 
spread of the S. typhimurium, and neutralize any released endotoxins that 
may be formed — thus allowing the remaining tissue to perform its vital 

Individuality of the Genotypes for Resistance to Different Diseases 

The question arises, is this condition one where the natural resistance 
extends to one or all diseases? does the individual have an over-all constitu- 
tion? or is the constitution specific for each disease? This question was 
investigated by Gowen and Schott (22) for diseases due to Salmonella 
typhimurium, pseudorabies and the antigenic poison ricin. The genes 
required for resistance or susceptibility to one disease were independent of 
those required for resistance to another. Webster (23) confirmed this 
observation for the independence of resistance to louping ill and to typhoid. 
For bacterial species that are related taxonomically, the natural resistance 
to one disease carries over to that due to its close relative. Typhoid 
resistance is closely correlated with Pasteurella resistance but less cor- 
related with Klebsiella or pneumococcus resistance. Similar conclusions 
may be derived from the data of Schutze, Gorer, and Finlayson (24) for 
S. typhimurium, S. enteriditis, louping ill, Pasteurella and pneumococcus. 

Taken broadly, the results concur in showing that a resistant constitu- 
tion for one disease is only likely to indicate resistance to another disease if 
the two diseases are fairly closely related. When the diseases are of differ- 
ent types, resistance to one tends to be independent of resistance to the 
others. In this respect inheritance for disease resistance behaves like any 
other inheritance dependent on many genes — some genes are independent; 
some appear to be linked or to have more than one effect; some have 
physiological interactions. But it would appear that all are ultimately 

Individuality of the Agent Initiating the Disease 

The bacteria or other agents (like the pollens of a given species) are 
just as or more diverse in their genotypes than the hosts on which they act. 
They may have more significance to the disease that is produced because 
they reproduce in such large numbers. The small rate at which mutations 
of existing genes occur and the rapidity with which such mutants may 
replace the existing population becomes of great importance in altering the 
character of the disease produced. The difference may result in changes 
in the growth pattern of the colony, in the color, morphology, or antigenic 
properties of the organism. They may affect the cultural requirement of 
the organism making some strains require a nutrient for growth which 
other strains can synthesize. The pathogenic characteristics may be 
altered in either the direction of greater or lesser virulence. The effects 
of these changes on disease expression have been under study in our 
laboratory for some 20 years for such diseases as tobacco mosaic, corn 

Journal of the National Cancer Institute 


wilt, and typhoids of both mouse and fowl. The results are concordant. 

Changes occur and may readily be established in all of the pathogens 
involved. The new types may appear at any time during the experiment 
and may cover a wide range. With selection it is possible to establish 
many of the new forms as true breeding types. In a given environment 
competition may exist resulting in rapid replacement of the unfavored 
types. Changes in virulence as dependent upon the inherited bacterial 
constitution have been examined by searching out phenotypic variants 
in originally pure stocks [Zelle {25), Lincoln and Gowen {26), Go wen 
{27), Plough, Young, and Grimm {28), Gowen, Stadler, Plough, and 
Miller {29)]. The mutant pheno types are represented by changes in color 
or morphology of the colonies, antigenic types of the organism, ability to 
synthesize amino acids or other metabolites from an energy source and 
simple salts. 

The pathogen's ability to initiate a disease in a host is rather highly 
specific. As attained in nature this property represents some chance 
combination of genes which appeared, was preserved, and improved upon 
by selection during successive generations of reproduction for better and 
better gene combinations and the inclusion of any of the more favorable 
mutations that might have occurred as time passed. Contact with an 
epidemic indicates that an efficient gene combination in the pathogen has 
evolved. Changes of a random sort comparable to those resulting from 
chance mutation would be unlikely to improve the disease-producing 
ability of a highly virulent organism. Our results agree with this inter- 
pretation. Mutations of our highly virulent lines on the average led to 
lines of less virulence. A study of 12 different mutants of 533-1 1C re- 
quiring adenine as a metabolite illustrate this fact {29). Seven mutants 
were avirulent causing no deaths, three showed some virulence, two were 
as virulent as the parent. The average virulence after mutation was 
less than the highly virulent parent type. The mutants were separated 
because of their effects on adenine. The range in virulence shows that 
this requirement is not itself the primary cause of virulence. Rather it is 
some change which through chance was associated with the change to 
adenine-requiring that is of importance. 

Individuality of the Host and Pathogen in Acquired Immunity 

Three factors important to increasing resistance to a disease through 
vaccination are common knowledge. The fourth factor and possibly the 
most important is not so generally recognized. In order of the emphasis 
given them, these factors are: 1) the dose of the vaccine administered at 
any one time; 2) the necessity for vaccinations at successive intervals; 
3) the line of the bacteria making the vaccine ; and 4) the genotype of the 
host receiving the vaccine. 

Clear evidence for the significance of the bacterial genotype in S. 
typhimurium immunization is found in our work [Gowen {SO)]. The 
typhimurium line which displays little virulence was the poorest im- 
munizer. The other two lines having higher virulence were also better 

Vol. 15, No. 3, December 19S4 

316263—54 18 

568 proceedings: symposium on 25 years of 

immunizers. The avirulent line and one of the virulent lines came from 
the third line by one-step and by two-step somatic segregation or muta- 
tions, respectively. As these bacterial lines come from each other by 
mutation, the importance of even a gene difference in the lines' genotypes 
is evident. 

The effects of the hosts' genotypes on their abilities to immunize like- 
wise show individuality [Go wen (80, 31)]. The 6 different strains im- 
munize differently. The naturally resistant strains on first contact with 
the disease are those which have their resistances enhanced most by im- 
munization. The intermediate strains, as measured by natural resist- 
ance, are likewise intermediate in their abilities to immunize. The most 
susceptible strains after immunization remain more susceptible than the 
other genotypes immunized in like manner. The level of resistance of 
each strain is simply raised a proportionate amount. In terms of bacteria 
which may be inoculated, immunized mice can resist 100 to 200 times as 
many live virulent-strain bacteria as the unvaccinated mice. It is un- 
fortunately true that the strains most badly in need of increased resistance 
through vaccination are the ones which remain most susceptible. 


The interplay of two sets of factors is important to determining the in- 
dividuality displayed by both host and pathogen in disease; those repre- 
sented by heredity and those of the environment affecting the expression 
of this inheritance. The inheritance mechanism has been utilized in the 
studies reviewed. From a population in which survival was a rare spo- 
radic event, it was possible to develop a population in which morbidity 
and death, instead, become the rare circumstance. Between these two ex- 
tremes, populations characterized by all grades of resistance and suscepti- 
bility were established. But little could be attributed to either passive 
or active humoral transfer of resistance from the parents of one generation 
to another. 

The character bases for this range in disease reaction, as portrayed by 
the correlations of resistance and susceptibility between inbred lines, give 
evidence of ability to resist weight change during the course of the disease ; 
heart, kidney, and liver size; blood volume and hematocrit percentages; 
particular serum albumins and globulins; leukocyte, chiefly lymphocyte, 
numbers; ability of the liver cells to isolate the disease and allow the re- 
maining cells to function in normal glycogen and fat metabolism; and 
ability of the macrophages to ingest and digest the pathogenic bacteria 
as factors of host constitution significant to the disease prognosis. On 
the other side, weight prior to attack seemed unimportant to natural 
resistance. Spleen size, despite its pronounced reaction to the disease, 
could not be considered indicative of the strain's resistance. Humoral 
elements in the form of agglutinins and precipitins were noticeable by their 
absence prior to infection. The humoral elements have their significance 
in acquired resistance. 

Journal of the National Cancer Institute 


Resistance or susceptibility is shown to be specific for the disease and 
host. A desirable constitution is an aggregation of many factors rather 
than an over-all constitution which is resistant or susceptible to all dis- 
eases. Many of these factors are hereditary, recombining and segrega- 
ting as is expected of characters under genetic control. 

Individuality in the pathogen may be generated by processes compa- 
rable to mutation. The breadth of the effects extend to varied attributes 
some of which are basic to life itself as invasive power of the organism or 
metabolism of particular proteins. Basic continuity of the inheritance 
makes the pathogen stable but mutations can occur, alter these characters, 
and create new forms. These mutations may change the population of 
organisms from virulent to avirulent or vice versa, or from a good immu- 
nizer to a poor, etc. 

The dual nature of disease reactions is further clarified in the template 
mechanism observed in acquired resistance. The genetic constitutions 
of host and of pathogen both require consideration. Susceptible host 
and avirulent pathogen genotypes are required for repeated attacks. 
The interactions involved in acquired immunity phenomena may explain 
many of the disappointments in this treatment approach. 

This search for elements which are significant to the expression of the 
typhimurium disease in mice emphasizes that no one element is crucial. 
Rather resistance is built upon the integration of many elements in the 
physiological well-being of the organism. 


(1) Schott, R. G.: The inheritance of resistance to Salmonella aertrycke in various 
strains of mice. Thesis, Iowa State Coll. Libr. 1-59, 1931. 

{2) : The inheritance of resistance to Salmonella aertrycke in various strains 

of mice. Genetics 17: 203-229, 1932. 

(3) Hetzer, H. O. : The genetic basis for resistance and susceptibility to Salmonella 

aertrycke in mice. Genetics 22: 264-283, 1937. 

(4) Lambert, W. V.: Genetic investigations of resistance and susceptibility to dis- 

ease in laboratory animals. Rep. Agr. Res., Iowa Agr. Expt. Sta. pp. 89-90, 
1931; 91-92, 1932; 115, 1933; 142-143, 1934; 158-159, 1935; 147-148, 1936. 

(5) Gowen, J. W.: Genetic investigations of resistance and susceptibility to disease 

in laboratory animals. Rept. Agr. Res., Iowa State Coll. Agr. Expt. Sta. pp. 
158-159, 1937; 151-153, 1938; 156-160, 1939; 192-194, 1940; 171-172, 1941; 
189-190, 1942; 178-182, 1943; 204r-210, 1944; 278-283, 1945; 257-260, 1946; 
230-232, 1947. 

(6) Lambert, W. V.: Breeding for resistance to fowl typhoid in poultry. Rept. Agr. 

Res., Iowa Agr. Expt. Sta. pp. 88-89, 1931; 114r-115, 1933; 142, 1934; 157-158, 
1935; 146-147, 1936. 

(7) : Natural resistance to disease in the chicken. I. The effect of selective 

breeding on natural resistance to fowl typhoid. II. Bacteriological studies upon 
surviving birds of the resistant stock in relation to progeny resistance. III. 
The comparative resistance of different breeds. J. Immunol. 23: 229-260, 1932. 

(8) Lambert, W. V., and Knox, C. W. : Mortality in chickens following the feeding 

of massive doses of virulent fowl typhoid bacteria. J. Am. Vet. M.A. (new 
ser.) 73: 480-483, 1928. 

(9) : The inheritance of resistance to fowl typhoid in chickens. Iowa State 

Coll. J. Sc. 2: 179-187, 1928. 

Vol. 15, No. 3, December 1954 

570 proceedings: SYMPOSIUM ON 25 YEARS of 

(10) : Selection for resistance to fowl typhoid in the chicken with reference to 

its inheritance. Iowa Agr. Expt. Sta. 153: 262-295, 1932. 

(11) Gowen, J. W., and Schott, R. G.: A genetic technique for differentiating between 

acquired and genetic immunity. Am. J. Hyg. 18: 688-694, 1933. 

(12) Grahn, D.: Estimation of the genetic influence on growth and organ weight 

changes in mice following total body X-irradiation. Thesis, Iowa State Coll. 
Libr., Ames, Iowa, 1952. 

(13) : Genetic implications of internal organ weight differences in inbred mice. 

Thesis, Iowa State Coll. Libr., Ames, Iowa, 1950. 

(14) : Genetic variation in the response of mice to total body X-irradiation. 

I. Body weight response of six inbred strains. J. Exper. Zool. 125: 39-62, 

(15) : Genetic variation in the response of mice to total body X-irradiation. 

II. Organ weight response of six inbred strains. J. Exper. Zool. 125: 63-83, 

(16) Thompson, S. : Serum proteins, leukocytes, and mortality of seven inbred mouse 

strains during cortisone administration and infection with Salmonella typhi- 
murium. Thesis, Iowa State Coll. Libr., Ames, Iowa, 1952. 

(17) Gowen, J. W., and Calhoun, M. L.: Factors affecting genetic resistance of mice 

to mouse typhoid. J. Infect. Dis. 73: 40-56, 1943. 

(18) Gowen, J. W., and Zelle, M. R.: Irradiation effects on genetic resistance of 

mice to mouse typhoid. J. Infect. Dis. 77: 85-91, 1945. 

(19) Weir, J. A. : The nature of genetic resistance to infection in mice. (Abstract.) 

Rec. Genet. Soc. America: 79, 1952. 

(20) Weir, J. A., Cooper, R. H., and Clark, R. D.: The nature of genetic resistance 

to infection in mice. Science 117: 328-330, 1953. 

(21) Oakberg, E. F.: Constitution of liver and spleen as a physical basis for genetic 

resistance to mouse typhoid. J. Infect. Dis. 78: 79-98, 1946. 

(22) Go wen, J. W., and Schott, R. G. : Genetic constitution in mice as differentiated 

by two diseases, pseudorabies and mouse typhoid. Am. J. Hyg. 18: 674-687, 

(23) Webster, L. T.: Inheritance of resistance of mice to enteric bacterial and 

neurotropic virus infections. J. Exper. Med. 65: 261-280, 1937. 

(24) Schutze, R., Gorer, P. A., and Finlayson, M. H.: The resistance of four 

mouse lines to bacterial infection. J. Hyg. (Cambridge) 36: 37-49, 1936. 

(25) Zelle, M. R. : Genetic constitutions of host and pathogen in mouse typhoid. 

J. Infect. Dis. 71: 131-152, 1942. 

(26) Lincoln, R. E., and Gowen, J. W. : Mutation of Phytomonas stewartii by X-ray 

irradiation. Genetics 27: 441-462, 1942. 

(27) Gowen, J. W.: Inheritance of immunity in animals. Ann. Rev. Microbiol. 2: 

215-254, 1948. 

(28) Plough, H. H., Young, H. N., and Grimm, M. R.: Penicillin-screened auxo- 

trophic mutations in Salmonella typhimurium and their relation to X-ray 
dosage. J. Bact. 60: 145-157, 1950. 

(29) Gowen, J. W., Stadler, J., Plough, H. H., and Miller, H. N.: Virulence and 

immunizing capacity of Salmonella typhimurium as related to mutations in 
metabolic requirements. Genetics 38: 531-549, 1953. 

(30) Gowen, J. W.: Genetic aspects of virulence in bacteria and viruses. Ann. 

Missouri Bot. Gard. 32: 187-211, 1945. 

(SI) : Humoral and cellular elements in natural and acquired resistance to 

typhoid. Am. J. Human Genet. 4: 285-302, 1952. 

Dr. George E. Jay, Jr., National Institutes of Health, Bethesda, Md. 

I would like to take this opportunity to express my thanks and appreciation to the 
Jackson Laboratory and to the arrangements committee for the invitation to partici- 
pate in this memorable symposium. It is certainly appropriate that 25 years of 
mammalian genetics and 25 years of the Jackson Laboratory be commemorated to- 
gether, for both have grown together. These past 25 years have been fruitful for both; 
may the next 25 years be just as bountiful. 

The role of a discussant has always been a puzzling one to me. I have asked various 
people about it and have received various answers, all of which contributed to my 
confusion. However, today at lunch some advice was offered which I believe I will 
follow. It was advised that a discussant should comment briefly on the paper (pro- 
foundly of course), raise some pertinent questions, and sit down. How profound my 
comments are will be questionable; I shall try to raise at least one pertinent question, 
and I assure you that I am real good at sitting down. 

A symposium is defined as either a drinking party or feast, or a conference at which a 
particular subject is discussed and opinions gathered. It seems to me that either def- 
inition may serve for this symposium, for certainly if this first paper is any indication, 
there will be ample opportunity to drink deep from the cup of knowledge and there will 
be an abundance of food for thought. For those of you more interested in the less 
intellectual aspect of drinking and feasting, the proposed extracurricular activities will 
provide such opportunities. As for the second definition, the subject of mammalian 
genetics is due for considerable discussion, and the expression of opinions will no doubt 
be made. 

The data presented by Dr. Gowen today are the results of a number of years of work 
by him and his associates, and to my mind it is a classic example of one aspect of re- 
search in mammalian genetics. Without a doubt it answers the original question 
posed: that heredity does play an important role in disease resistance and/or suscepti- 
bility. Just how hereditary factors operate is still not clear, though Dr. Gowen's data 
offer some interesting leads. The differences between the strains inbred for various 
levels of resistance show that many physiologic and morphologic manifestations enter 
into the creation of any particular level of resistance. 

These manifestations that characterize the various inbred strains are actually the 
genetic results of individuality expressed in the original heterogeneous population. 
This individuality was used to synthesize new groups of animals, each differing from 
the other and expressing an individuality characteristic for that group or inbred strain. 
The genetic technique of close inbreeding to fix such individuality is one that has 
become a mainstay in modern mammalian genetics, and is a technique perhaps 
synonymous with the Jackson Laboratory. Thus it is indeed fitting that Dr. Gowen's 
paper is first on the schedule, for it properly illustrates the employment of this tech- 
nique in sorting out the factors contributing to disease resistance. 

These ideas, of course, have long existed in the field of cancer research. In fact, it 
was in this research area that Dr. Little first applied these ideas and practices. Since 
most of you present are well acquainted with the genetic concepts of cancer research, 
this area will only be mentioned. Within the past few years, genetic concepts and 
methods are being used in still other areas of research. In dental caries, Hunt and his 
associates have been able to develop two strains of rats, one resistant and the other 
susceptible to carious lesions. As a result, material is now available in limited quanti- 
ties which will permit more quantitative studies on the factors involved in the elabora- 
tion of caries. 

In certain aspects of nutrition research, the individuality of inbred strains is coming 
into its own. Some recent work at the National Institutes of Health by Dr. Klaus 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 


Schwarz has shown that one strain of inbred rats is particularly suitable for studies on 
dietary liver necrosis. He has found this strain (Fischer 344) to be uniformly suscep- 
tible to dietary liver necrosis, and thus it probably will be good material for testing 
unknown dietary preparations in research of this nature. 

Recent studies by Dr. E. S. Russell and her associates, on differences between inbred 
strains of mice in blood constituents, offer interesting material for basic research, as well 
as possible testing systems for biologic preparations. Some preliminary work on drug 
metabolism by me indicates marked strain differences, which may have value as far as 
drug testing is concerned, and certainly offers possibilities for basic studies on metabolic 
pathways. These and other examples can be cited as new illustrations of the uses and 
importance of individuality as expressed between inbred strains in future medical 

Thus the work of Dr. Gowen, the work in cancer, and the more recent work in the 
other areas just mentioned, are all examples of the control and utilization of individu- 
ality of inbred strains in research. The question I now raise is this: Is it not possible 
to carry this general concept a step further and utilize the individuality as expressed 
by individuals within these inbred strains? As you know, an inbred strain is highly 
uniform, for theoretically most of the individuals within the strain have the same or 
nearly the same gene complement. However there occurs at intervals (sometimes 
more often than the geneticist cares to admit) , individuals within a strain that are not 
like the others. They may differ in morphologic or physiologic aspects to such an 
extent that they are obviously different — they exhibit individuality. This individu- 
ality may be the result of mutational changes, recombinations, or some other phe- 
nomenon. But whatever its cause, it is a situation that may be of considerable value. 
For example, in both the C57BL/6 and C57BL/10 strains of mice, Dr. Paulo Borges has 
observed and reported changes, found to be mutations from one histocompatibility 
gene locus to another. These mutations resulted in profound differences in the 
responses to certain transplanted tumors. Thus, the sublines which now exist for each 
of these two parent strains probably differ by only one gene from the parent strain. 
This situation seems to offer a unique opportunity for studies in the action of histo- 
compatibility genes. No doubt many other differences similar to this one have been 
observed, and perhaps many of you have speculated on the possibilities these differences 
offer. It would seem entirely reasonable to think that such expressions of individuality 
within an inbred strain may well provide additional clues for further differentiating 
the factors that contribute to the final manifestation of a given characteristic. Dr. 
Gowen's continuing work with the inbreeding for differences in leukocyte numbers is 
actually an elaboration of this point. 

Just how far one can go with the control and utilization of individuality is yet to be 
seen. The creation of inbred strains was the first step, the elaboration of sublines of 
inbred strains perhaps the second, and the selection out of individual differences from 
these larger groups the third. The future offers some interesting possibilities for 
further genetic manipulations. 

The Importance of Differences in Mor- 
phology in Inbred Strains * 

Thelma B. Dunn, Laboratory oj Pathology , 
National Cancer Institute, 2 Bethesda, Md. 

When Dr. Russell invited me to this symposium, I wrote her that there 
was no one whom I would more delight in seeing honored than Dr. Eliza- 
beth Fekete, and no subject in which I have a more enthusiastic interest 
than the one she suggested for me. Dr. Fekete has contributed greatly to 
our knowledge of the morphology of the inbred strains. Whenever a 
new pathologist joins our staff, we immediately introduce him to the 
chapter by Dr. Fekete on Histology, which appears in that much prized 
contribution from this laboratory, The Biology oj the Laboratory Mouse. 
This chapter is like a compass to the pathologist trained in human anatomy 
who must orient himself in the field of cancer research. The intricacies 
of morphology in the mouse are clearly and accurately described by one 
who knows firsthand. After this introduction, one can then proceed 
with the many other contributions which Dr. Fekete has made to mor- 
phology in her studies of inbred strains. 

All morphologists should be grateful to the laboratory here and its 
staff for the development of inbred strains of mice with their comforting 
uniformity. While we may encourage individuality for the best operation 
of a democratic society, conformity simplifies biologic experiments. The 
first requirement in animal research is a reliable group of controls. This 
is especially true in cancer studies, since observations must often be made 
on animals at an advanced age, when the unequal onslaughts of time are 
added to inborn variations. Inbred strains are our greatest aid, and 
while noninbred mice may be tolerated for some studies when young 
animals only are used, they are not acceptable for studies of cancer. 

All the morphologist can be sure of is what he sees on gross examination 
or in a microscopic slide. The microscopic slide represents a single 
instant in a long series of events. What happens in the intervals before 
and after is a blank, and must be filled in by conjecture. Since even the 
direction of movement is sometimes uncertain anything which regulates 
the order of these still-life views is a boon. Most pathologists will agree 
as to what is actually there, but in filling up the intermediate blanks, they 
do not always see eye to eye. 

» Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 27, 1954. 
2 National Institutes of Health, Public Health Service, U. S. Department of Health, Education, and Welfare, 


Journal of the National Cancer Institute, Vol. 15, No. 3, Deeember 19S4 

574 proceedings: symposium on 25 years of 

Inbred mice furnish a group of nearly identical individuals progressing 
toward senility on a fixed schedule. Such morphologic and chronologic 
uniformity is not to be found in random-bred animals. An individual 
removed from a heterogeneous group of mice may be a variant, or notably 
precocious or laggard in reference to the others. 

Inbred strains give animal experiments a notable advantage in com- 
parison with observations on human beings. In clinical research a valiant 
effort is made to get as uniform a group as possible. Except for a limited 
supply of identical twins, healthy male medical students, in their early 
twenties and of obliging disposition, probably represent the closest human 
approach to an inbred strain. A clinician does not walk out on the street 
and pick up the first 100 passersby to determine such a character as 
normal blood pressure. Yet I was amazed to find a pathologist looking 
for a group of ordinary house mice which he regarded as a ' 'basic" or 
"universal" mouse, for he assumed that inbred strains represented abnor- 
mal and specialized types. 

The importance of inbred strains in the identification of genetic mech- 
anisms, in the recognition of factors related to bacteriologic resistance or 
susceptibility, and to investigations in many other fields will be con- 
sidered by other speakers. 

My discussion of the morphologic difference in inbred strains has been 
arranged under four headings. These are: 1) Development in early life, 
and the emergence of pathologic lesions as the animal ages, 2) comparison 
of morphologic features and their relation to other factors, 3) incidence 
of neoplasms, and 4) recognition of precancerous conditions or changes. 
Many examples will overlap and the purpose of this grouping is con- 
venience rather than exactness. My observations are based largely on 
my own experience; much of this material is unpublished. Spontaneous 
diseases have usually been selected as illustrations, for the number of 
induced lesions in which morphologic differences are important is too 
extensive to be reviewed here. Such a wealth of illustrative material to 
emphasize the desirability of inbred strains is at hand, that the real 
difficulty lies in selecting examples. 

1. Development in Early Life, and Pathologic Changes with Age 

The first example showing strain differences in early life illustrates the 
precision, the exact timing, which inbred strains exhibit. Because of 
this adherence to chronology, inbred strains should always be used in 
studies of development. Dr. James B. Longley (1) of the National 
Institutes of Health was looking for a morphologic difference in portions 
of the proximal convoluted tubule in the kidney, where it is known that 
functional differences exist. He determined by special stains that alka- 
line phosphatase disappeared entirely from the more distal portion of the 
tubule as the kidney of the mouse matured. Animals were then selected 
at frequent intervals from birth to maturity, and estimations were made 
of the amount of alkaline phosphatase in different segments of the tubule. 

Journal of the National Cancer Institute 


The time of disappearance was found to be remarkably uniform in indi- 
viduals from the same inbred strains, but differed in individuals from 
different strains. The periodic acid-Schiff technique also showed altera- 
tions as the kidney matured. Sections of the kidney from strain DBA 
mice, age 29 days, still contained alkaline phosphatase in the more distal 
section of the proximal tubule. At 31 days the alkaline phosphatase 
reaction was fading, and at 36 days, it had almost completely disappeared. 
Graphs to show the estimated quantity of alkaline phosphatase in the 
kidney during this maturation period were prepared, using several inbred 
strains. The disappearance was gradual in strains C3H and C57BL. 
In strains BALB/c and DBA, disappearance began a few days later, then 
progressed rapidly. When noninbred mice were tested, the curve was 
diphasic, implying a decline, then an increase, in the enzyme. It is certain, 
however, that once alkaline phosphatase disappeared it did not reappear, 
so a false premise might have been reached if observations had been 
confined to the noninbred mice. 

Senile degenerative changes are as much a consequence of time as are 
the developmental changes of early life. A kidney lesion of mice found in 
late life is due to amyloid deposition. Kidneys of old mice, especially 
of strain A, often show scarring of the surface and many small cysts. 
Study of the kidneys alone failed to reveal a cause for this alteration, 
but other organs in the same mice showed amyloid deposition (#). Dr. 
Heston and Dr. Deringer (8) crossed strain A mice with C57L, a strain 
in which amyloidosis is infrequent, and examined the hybrids and back- 
crosses for amyloid. In strain A, the incidence was 88 percent. In strain 
C57L the incidence was 9 percent. The first generation hybrids between 
these strains had an incidence of 5 percent. When backcrossed to strain 
A, the incidence was 49 percent, and when backcrossed to strain C57L 
the incidence was 8 percent. When amyloidosis appeared in strain C57L 
it was secondary to dermatitis. Primary amyloidosis and the accompany- 
ing renal damage thus occurred in incidences conformable with the 
inheritance of a single recessive gene. 

The type of kidney damage secondary to amyloidosis apparently de- 
pends upon the site of deposition in the kidney. This varies with different 
strains. In strain A the amyloidosis was primary or idiopathic and af- 
fected many organs. Amyloid often was not found in the kidney at a late 
stage, but was found at an earlier stage deposited in a transverse line 
across the papilla (4). This led to necrosis and sloughing of the papilla. 
Atrophy and cyst-formation in the cortex were secondary to obstruction 
of the collecting tubules. In strain C57BL, the amyloid was deposited 
in the glomeruli, the tubules were remarkably well preserved, and the 
papilla was intact. In strain HR, amyloid deposition was frequent 
in the glomeruli and there was also calcification of the tip of the papilla, 
and sloughing. 

These observations on the differences in amyloid deposition in inbred 
strains have usually been made on random groups. Except for strain A 
and C57L no large series has ever been tabulated. A careful study on 

Vol. 15, No. 3, December 1954 

576 proceedings: symposium on 25 years of 

many animals from other strains might definitely establish characteristic 
strain differences in amyloid deposition. 

A kidney lesion showing remarkable strain specificity was produced in 
strain A after repeated injection or ingestion of urethan (5), given to 
produce lung tumors. About half of the animals developed ascites. On 
microscopic examination a glomerular lesion was found which had many 
of the morphologic features of glomerulonephritis in man. An experi- 
mental disease of this type should be helpful in studying renal disease, for 
there are no good counterparts of human glomerulonephritis in animals. 
With this in mind urethan was given to several other strains of mice, and 
to rats, rabbits, and guinea pigs. All our attempts to produce kidney 
damage in other strains or species were unsuccessful. Kirschbaum and 
Bell (6) later reported a similar lesion in strain NHO mice in which a 
spontaneous disease of similar type had previously been described. 

Another example of a complex process manifested by morphologic 
differences in the kidney was produced by the exposure of a number of 
mice of inbred strains to chloroform (7) . The males were more seriously 
affected than females, and mice of some strains more than others. Nearly 
all strain C3H males were killed by a dose which nearly all females of 
this strain survived. When males survived, the kidney cortex usually 
showed calcification when the mice were autopsied several months later. 

These observations on the kidney demonstrate that when routine 
histologic methods are applied to the kidneys of normal mice, no strain 
differences may be detected. By the use of special stains on the kidney 
during development, however, or from the study of a degenerative con- 
dition as amyloidosis, or following the exposure to a toxic substance, 
morphologic differences in the strains are readily shown. 

Let me now recount a series of adventures, or misadventures, where 
inbred strains led me to believe I had made sensational discoveries. In 
spite of this deception, I am still grateful, for it was the presence of the 
same lesions in control mice from inbred strains which corrected the 
errors before there was time for publication. 

Dr. Harold Morris (8) fed 5 inbred strains of mice synthetic diets in 
which pyridoxine was absent or deficient. The experiment became 
exceedingly involved, for each inbred strain reacted differently. The 
effect on survival time varied with the strain. With chronic deficiency, 
when a small but inadequate amount of pyridoxine was given, some mice 
of strains C58 and C57BL survived about 56 weeks while all strain DBA 
mice were dead by 38 weeks. 

Lesions related to the deficiency depended somewhat upon the length 
of time the deficient diet was given, and varied in different strains. Strain 
DBA showed severe anemia and damage to the reproductive organs. 
With chronic deficiency, C57BL developed a dermatitis not seen in 
other strains. Some C58 mice had a locomotor disturbance. When not 
severely deficient, a few strain A mice lived even longer than the controls, 
for they did not develop the characteristic strain A amyloidosis. If one 
reasoned from the particular to the general, and considered the results 

Journal of the National Cancer Institute 


in strain A mice only, it might be argued that a mild pyridoxine defi- 
ciency is desirable for long life. A remarkable skeletal lesion developed in 
strain C58 mice in the deficient group. At first it was thought to be a 
hitherto undisclosed effect of pyridoxine deficiency, but when the control 
group on the synthetic diet was examined the same disturbance was 
found in them. Osteoid tissue developed about the shafts of the rib 
(fig. 1), and many other locations in the skeleton, especially at sites sub- 
ject to trauma. It appeared in 0.83 percent of old strain C58, which had 
been for a long period on the synthetic diet and in 43 percent of strain 
C57BL, which is related to strain C58. It was found in 5 percent of strain 
DBA and never in strains C3H and A, which are related to each other but 
unrelated to strains C58 and C57BL. We have not observed such a 
lesion in our mice on a regular diet. 

After noting the bone lesion the next obvious step was to examine the 
condition of the parathyroid glands. It was found that in many of the 
mice with the bone lesions, the parathyroid glands were dark in color (9). 
The association with the bone lesion appeared significant until the same 
pigmentation was found in untreated C58 mice. 

When the entire parathyroid gland was cleared in glycerine, and when 
microscopic sections were examined, it was evident that the dark color on 
gross examination was due to melanocytes in the parathyroid stroma. 
Similar dendritic melanocytes have been found in the heart valves and the 
meninges of strain C58. This peculiarity might be used by embryologists 
to follow the spread of melanoblasts from the neural crest. 

The melanocytes made the parathyroid gland easily visible, so that 
parathyroidectomies could be performed. The mice survived this opera- 
tion — too well for our purpose — since none of them seemed any the worse 
for the experience. Still hoping to detect some endocrine effect, para- 
thyroid tissue from a number of mice was injected into others of strain 
C58, but they seemed as unmoved by a superfluity of parathyroid tissue 
as they were by deprivation. Nevertheless, complete endocrine examina- 
tions were made at autopsy. The pituitary gland in an occasional mouse 
was found to be cystic and several times the normal size (fig. 2). The 
pituitary glands of many untreated C58 mice were then examined, and 
although pituitary cysts were rarely found, just once was often enough to 
disprove a significant relation to the parathyroid. The lesion was not 
frequent enough, however, for a controlled experimental study. 

2. Comparison of Morphologic Features in Inbred Strains 

Another personal experience will be given as an illustration of this 
section. Year-old strain A mice had received 400 r irradiation at birth. 
Damage to the lens and severe damage to the retina were regularly found 
(10) (fig. 3C). The slides were shown to others in our laboratory. A few 
days later, another pathologist found a similar alteration in the retina of 
a strain C3H mouse which had never been exposed to X ray.||This was 
disconcerting. I consulted Dr. Heston, as I often do. He gave me a refer- 

Vol. 15, No. 3, December 1954 

578 proceedings: symposium on 25 years of 

ence to retinal anomalies in mice, which explained the puzzle. We now 
observe that all our strain C3H mice at Bethesda.have a secondary degen- 
eration of the rod layer in the retina (11). We found the retina to be 
normal at 10 days of age (fig. 3A), but degenerated by the time the mouse 
reached 28 days (fig. 3B). Sight must not be very important to a labora- 
tory mouse. We had not suspected before that all our strain C3H mice 
are blind. Fortunately, for the validity of the irradiation experiment, 
this anomaly has not been found in strain A mice. Ascribing the retinal 
damage to irradiation in strain A was justified, and we did not have the 
difficult and uncertain task of separating the induced lesion from a spon- 
taneous retinal anomaly. 

There is no organ which should be so lacking in distinction as the spleen. 
In the mouse the spleen appears to be an annex organ, an overflow tissue 
where the business which cannot get done in the bone marrow or lym- 
phatic tissue is carried on. One would think a spleen in any mouse should 
be just a spleen, and nothing more. Yet morphologic differences in spleens 
from various strains of inbred mice are sufficiently distinctive so that a 
shrewd guess as to the strain of origin can often be made by an examination 
of the spleen alone. While examining a group of strain A mice which 
had been treated with urethan for a prolonged period, an astonishing 
number of mast cells was observed, and it was supposed that urethan 
might be responsible. The detection of any drug or condition which 
affects the number of mast cells would be a real achievement. However, 
when the controls were examined, mast cells were equally numerous. 
We next counted the mast cells in cross sections of the spleen taken from 
various strains of mice at different ages and stained with toluidine blue 
{12). We found that the mast cell content of the spleen is a fairly reliable 
strain characteristic. Strain A had the highest number. In old males 
the number was often above 2,000 and could not be estimated accurately. 
Strain C57L had a smaller number averaging about 255 in males, and 16 
in females, and the number in the hybrids of these strains lay somewhere 
between the parents. Strain I had almost no mast cells. The number 
generally increased with age and was higher in males. 

Mast cells cannot be accurately enumerated in any other organ, so it 
is uncertain whether the total number of mast cells in the entire body of a 
strain A mouse greatly exceeds the total number in other strains. Old 
hybrids with a strain A parent have had more mast cell neoplasms than 
any other mice in our laboratory. The function of the mast cell has been 
the object of an intensive search in the past few years. An investigation 
of the differences in inbred strains might furnish a clue. 

A more subtle, but equally dependable difference in spleens from dif- 
ferent strains was found by Dr. Oscar Duque, working in the Pathology 
Laboratory at the National Cancer Institute (18). Dr. Duque is from 
Latin America, and perhaps it takes Spanish blood to do good silver 
stains. He made beautiful preparations, using a modified del Rio Hortega 
method, so that he could visualize the reticulum cells in the mouse spleen. 
A striking difference in the arrangement and quantity of the cells was 

Journal of the National Cancer Institute 



found in inbred strains. For example, a perifollicular collar formed of 
reticulum cells is well developed in strain I, but generally absent in strain 
C58. Reticulum cells are very numerous in the red pulp in strain I. In 
strain C58, the follicles are small and numerous. Within the follicles, 
the reticulum cells are prominent; those in the germinal centers being 
very large. A difference in the number and arrangement of reticulum 
cells is also shown if stains for iron are done on spleens from old mice. 
The reticulum cells are actively phagocytic for hemosiderin, which is 
readily identified by the Prussian-blue reaction, and the different strains 
show some of the same differences in reticulum that is seen after silver 

I have also observed that strain C57L has little hematopoietic activity 
in comparison with other strains. These comparisons might be thought 
a pastime of academic interest only, except that the modification of X-ray 
damage which results from shielding the spleen varies in different inbred 
strains. It is possible that morphologic differences in the strains may 
explain these differences and indicate the tissue or cell which is most 
important. The painstaking work from Dr. Russell's laboratory (14) on 
differences in the peripheral blood of inbred strains also reveals the slight 
but positive differences which must exist in the hematopoietic and lym- 
phatic tissues of inbred mice. 

In studies of another system Dr. Fekete described differences in the 
ovaries from inbred strains which correlate, to some extent, with functional 
endocrine activity (15). 

Next I will list briefly a number of degenerative lesions which are pecul- 
iar to some inbred strains. Attention is called to them, first, as a warning 
against supposing that they are produced by an experimental procedure, 
and, second, to invite you to consider a possible relationship to physiologic 
or metabolic differences among the strains. 

a) Hemosiderin is abundant in the reproductive organs of old females 
of strain BALB/c. The uterine wall contains large amounts of material 
stained by the Prussian-blue reaction, and the epithelial cells of the mam- 
mary ducts are packed with hemosiderin. This degree of hemosiderosis 
has not been observed in other strains. Can it be correlated with any 
other peculiarity in this strain? 

b) A corneal lesion has been frequently observed in strain DBA. This 
appears in the living mouse as opaque patches in the cornea. It was first 
observed in experimental mice, and was suspected of being an induced 
lesion (16). A degenerated, often calcified area in the myocardium is also 
commonly found in strain DBA, and in old C3H and C3Hf. Are these 
lesions secondary to a metabolic abnormality of these strains? 

c) The name ' 'mesenteric disease' ' was applied to a lesion of the mesen- 
teric node by Simonds (17). It is common in old strain C3H mice, and 
is found in lower incidence in strain C3H hybrids. An extreme enlarge- 
ment of the node is found, and microscopically there are many blood-filled 

Vol. 15, No. 3, December 1954 

580 proceedings: SYMPOSIUM ON 25 YEARS of 

d) A lesion of the pancreas appeared a number of times in a group of 
old C3Hf mice from Dr. Heston's colony which I examined recently. 
Cysts were formed by dilatations of pancreatic ducts. Islet tissue persisted, 
but all the acinar tissue disappeared in large areas. Occasionally these 
cysts ruptured and a condition of fat necrosis, such as is found in human 
pancreatic disease appeared. 

3. Neoplasms in Inbred Mice 

When I first began working with mice, I always consulted an experienced 
"mouse" pathologist regarding unusual autopsy specimens and slides. 
Before he would give me advice or make a diagnosis, he asked "what 
strain?" Depending upon my answer, the finding was considered either 
commonplace or remarkable. It took me some time to realize the impor- 
tance of this question. I now find that it is the first which comes to mind 
when I am shown any lesion or neoplasm in a mouse. Few features are 
more characteristic of an inbred strain than its tumors. Many examples 
of this are familiar to you. If mammary tumors in strain C3H, lung 
tumors in strain A, and leukemia in strains C58 and ML were suddenly 
taken away a good percentage of cancer research would be wiped out. 
In many laboratories these tumors are classified by site only, but close 
histologic study of large groups of tumors might reveal that even among 
these much investigated subversives, differences in morphology may exist 
in inbred strains. For example, in mammary tumors from several inbred 
strains, various types of histologic structures were found to be preponder- 
ant when the milk agent had been removed. In Dr. Heston's C3H mice 
without the agent, adenoacanthomas were common (18). In Dr. Ander- 
vont's DBA without the agent (19), but treated with methylcholanthrene, 
carcinosarcomas were frequent and Dr. Muhlbock (20) working with 
agent-free DBA found a high percentage of "deviating types" of mammary 
tumors many with marked squamous metaplasia. We have described a 
distinctive type of mammary tumor that most often appears in hybrid 
mice of advanced age, usually without the milk agent (21). We have 
termed this adenocarcinoma type C. This tumor is composed of numerous 
small cysts lined by a cuboidal epithelium, which is closely invested with 
a fusiform cell. Unlike most other mammary tumors, this type is generally 
uniform throughout the section. A van Gieson stain shows that the fusi- 
form cell is of muscle origin and not fibroblastic, for no collagen is devel- 
oped. This is the most distinctive characteristic of this tumor. When 
Dr. Foulds from England was here, we compared our experiences with 
mammary tumors in mice. He told me he had never seen the type just 
mentioned, although he had observed another pattern combining an 
epithelial and a muscle element. He showed me other types which I had 
not observed in any strains used in our laboratory. These differences 
stress the need for caution in transferring information regarding tumors 
from one strain to another. A standard mammary tumor is not to be 
expected in all strains. One can say only that certain morphologic 

Journal of the National Cancer Institute 


characters were found in the examination of tumors from a specified strain 
under specified conditions. 

Whether other common neoplasms, such as lymphocytic leukemia or 
pulmonary tumors, might also show slight variations in different strains 
of mice has not been determined. 

Many remarkable tumors were disclosed in a survey made by Dr. 
Heston and Dr. Deringer, which Dr. Heston will describe more fully. 
Strain C3H females with the milk agent were crossed with C57BL males, 
and then backcrossed to C57BL males through seven generations, until 
the mice were largely of strain C57BL constitution. Then by a reverse 
process, the females of this seventh backcross generation were crossed 
with strain C3H males, without the agent, and the resulting females were 
backcrossed to C3Hf males through four generations until a nearly pure 
C3Hf mouse resulted. The following rare tumors appeared, and were 
rather definitely segregated according to genetic constitution. 

a) Osteogenic sarcomas. — In strain C3Hf and in hybrids without the 
agent derived from C3H or C3Hf . The morphology is being described by 
Dr. Hilberg of the National Cancer Institute. A pattern of interlacing 
cords, with ossification at the center is characteristic (22) . 

b) Harderian gland tumors. — Most frequently in C3Hf and in hybrids 
without the agent derived from C3H or C3Hf . Neoplastic tissue replaces 
the Harderian gland, but structural features of the normal gland are 
retained. There was considerable variation in the degree of differentiation 
in different tumors. 

The next examples show how the incidence of tumors characteristic of 
an inbred strain may be altered. The agent had been removed from strain 
C3H, and the mammary-tumor incidence was reduced so that many mice 
survived to old age. The following rare types of neoplasm then emerged. 

c) Adenocarcinomas of the uterus. — Many of the adenocarcinomas were 
polypoid, and closely resembled adenocarcinoma of the uterus in women. 
I have not found them previously described. 

d) Ovarian tumors. — Eare types of ovarian tumors were found in strain 
C3Hf. One, a papillary cyst-adenocarcinoma was previously described 
from the laboratory here (23) and among noninbred mice of Maude Slye's 
colony (24) • Another very rare type produces a mucoid substance, and 
somewhat resembles the mucoid neoplasms of the intestine in man. One 
of these mucoid tumors has been transplanted successfully. It grows 
slowly, it maintains the same morphology and produces a mucoid sub- 
stance as did the original. I have not found a previous description of this 

Many other examples of tumors peculiar to one strain might be men- 
tioned. We have found localized plasma-cell tumors only in strain C3H 
(25). The majority of our mast-cell neoplasms have appeared in hybrids, 
derived from outcrossing strain A. Tumors of the central nervous system 
are described only in strain NHO (26). 

It thus appears that an entirely pathologic and destructive process 

Vol. 15, No. 3, December 1954 

582 proceedings: symposium on 25 years of 

a neoplasm, may be one of the most characteristic features of an inbred 

4. Recognition of Precancerous Conditions or Changes 

This involves the most difficult and uncertain of all our endeavors. 
Investigators working with mice have perhaps been overly eager in 
attempts to correlate some newly discovered morphologic feature or 
physiologic difference in a strain with cancer. It is hopefully assumed 
that a victim destined for cancer carries with him some distinguishing 
mark, which separates him from others. De Quincey remarking upon 
the frequency of tuberculosis among his countrymen said that one destined 
for this disease carried a sign, like a tree which was blazed for cutting. 
In searching for such a mark in cancer, the fallacy often lies in assuming 
that there is a general tendency to cancer no matter at what site it develops. 
Cancer, however, in mice and probably in man, is a local disease of a 
specific organ. A high-mammary tumor strain may be a low-leukemia 
strain. A morphologic or physiologic variation promoting cancer should 
therefore reside in or be closely related in some fashion with that particular 
organ or tissue in which the disease is found. Several errors of correla- 
tion, or unsuccessful efforts at correlation, to mammary tumor develop- 
ment may be mentioned, such as brown degeneration of the adrenal (27), 
the porphyrin content of the Harderian gland (28), differences in blood 
groups (29), and in iron content of the mammary tissue (80). Despite 
these failures, the discovery of precancerous conditions is among our most 
important endeavors. Recognition of them would give the experimentalist 
a clue to etiology, and the clinician a chance at prevention and early 
treatment. Many examples of precancerous conditions in clinical medi- 
cine are available, such a polyposis of the colon, xeroderma pigmentosa, 
and hereditary multiple skeletal exostoses. These are all local conditions 
preceding cancer of the same site. When the search in mice is directed 
toward a specific organ or tissue in which the cancer will later develop, 
some positive correlations have been found. We have for example, the 
hyperplastic nodule which precedes many mammary tumors and is much 
more frequent in mice with the milk agent (31), and the pathologic altera- 
tion in the skin of hairless mice (82), in which a high incidence of spon- 
taneous skin cancer develops. Examples of preneoplastic lesions are 
numerous if carcinogenic agents are used. On the other hand, a number 
of lesions have been observed where the histologic structure strongly 
suggested that they might terminate in neoplasia. Even with the 
stimulus of a carcinogen, many of these have proceeded no further than 
hyperplasia. For example, a gastric lesion in strain I mice proliferates 
wildly, but it does not progress to gastric cancer (88). Leukemoid 
reactions, with granulocyte counts in the hundreds of thousands, have 
never been successfully transplanted (84). A lesion in strain C58 was 
described as preleukemic (85), but this also appears in low leukemia 
strains in our laboratory. It was noted in the Jackson Laboratory that 
leukocyte counts in young mice could not be correlated with a later devel- 

Journal of the National Cancer Institute 


opment of leukemia in various inbred strains (36). In our laboratory 
lymphatic and blood-forming organs in high- and low-leukemia strains of 
mice at 6 weeks and 12 months of age were examined. Many interesting 
differences were observed but none of them appeared to be related to the 
development of neoplasms. Altogether, this line of research has been 
disappointing. We have usually failed to discover an abnormality in 
early life foretelling the later appearance of cancer. Regardless of these 
failures, inbred mice still offer our greatest hope in this search. In high 
tumor strains we have individuals that will develop cancer of a particular 
type at a particular site within a short period of time. In a corresponding 
low tumor strain the individual will not develop cancer in the particular 
organ within its normal life span. The significant difference in these mice, 
however, has usually been too deeply hidden for the morphologist to 
discover. Nevertheless, this is a line of research which must be continued 
and, as always, with inbred strains. 

In closing let me say that like Emily Dickinson's bee and honey the 
pedigree of my human associates does not concern me in the least, but I 
am very particular, even snobbish, about having the family history of any 
mouse that I am asked to work with. 


(1) Longlet, J. B., and Fisher, E. R.: Alkaline phosphatase and periodic acid- 

Schiff reaction in proximal tubule of vertebrate kidney. A study in segmental 
differentiation. Anat. Rec. 120: 1954. In press. 

(2) Dunn, T. B.: Relationship of amyloid infiltration and renal disease in mice. 

J. Nat. Cancer Inst. 5: 17-28, 1944. 

(3) Heston, W. E., and Deringer, M. K.: Hereditary renal disease and amyloidosis 

in mice. Arch. Path. 46: 49-58, 1948. 

(4) Dunn, T. B.: Some observations on the normal and pathologic anatomy of the 

kidney of the mouse. J. Nat. Cancer Inst. 9: 285-301, 1949. 

(5) Dunn, T. B., and Larsen, C. D.: Hyalinization of glomeruli produced in strain 

A mice by the administration of urethan (ethyl carbamate). (Abstract.) 
Fed. Proc. 5: 220, 1946. 

(6) Kirschbaum, A., and Bell, E. T.: Induction of renal glomerular lesions by 

urethan in inbred mice susceptible to spontaneous glomerulonephritis. Proc. 
Soc. Exper. Biol. & Med. 64: 71-72, 1947. 

(7) Deringer, M. K., Dunn, T. B., and Heston, W. E.: Results of exposure of 

strain C3H mice to chloroform. Proc. Soc. Exper. Biol. & Med. 83: 474r-479, 

(8) Morris, H. P., and Dunn, T. B.: Pyridoxine deficiency. Nutritional and 

histological observations on five strains of mice. 12th Congr. Pure and Applied 
Chem. 13: 164-165, 1951. 

(9) Dunn, T. B.: Melanoblasts in the stroma of the parathyroid glands of strain 

C58 mice. J. Nat. Cancer Inst. 10: 725-733, 1949. 

(10) Lorenz, E., and Dunn, T. B.: Ocular lesions induced by acute exposure of the 

whole body of newborn mice to roentgen radiation. Arch. Ophth. 43: 742-749 

(11) Tansley, K.: Hereditary degeneration of the mouse retina. Brit. J. Ophth. 

35: 573-582, 1951. 

(12) Deringer, M. K., and Dunn, T. B.: Mast-cell neoplasia in mice. J. Nat. 

Cancer Inst. 7: 289-298, 1947. 
(IS) Duqu£, O.: Unpublished. 

Vol. 15, No. 3, December 1954 
316263—54 19 

584 proceedings: symposium on 25 years of 

(14) Russell, E. S., Neufeld, E. F., and Higgins, C. T.: Comparison of normal 

blood picture of young adults from 18 inbred strains of mice. Proc. Soc. 
Exper. Biol. & Med. 78: 761-766, 1951. 

(15) Fekete, E.: A morphological study of the ovaries of virgin mice of eight inbred 

strains showing quantitative differences in their hormone producing compo- 
nents. Anat. Rec. 117: 93-113, 1953. 

(16) Morris, H. P., Dunn, T. B., and Wagner, B. P.: Influence of gonadotrophin 

on pyridoxine-deficient and diet-restricted female mice. J. Nat. Cancer Inst. 
14: 493-511, 1953. 

(17) Simonds, J. P.: Leukemia, pseudoleukemia, and related conditions in the Slye 

stock of mice. J. Cancer Res. 9: 329-373, 1925. 

(18) Heston, W. E., Deringer, M. K., Dunn, T. B., and Levillain, W. D.: Factors 

in the development of spontaneous mammary gland tumors in agent-free 
strain C3H b mice. J. Nat. Cancer Inst. 10: 1139-1155, 1950. 

(19) Andervont, H. B., and Dunn, T. B.: Response of mammary-tumor-agent-free 

strain DBA female mice to percutaneous application of methylcholanthrene. 
J. Nat. Cancer Inst. 10: 895-925, 1950. 

(20) Muhlbock, O., Tengbergen, W. van E., and Rijssel, Th. G. van: Studies on 

the development of mammary tumors in dilute-brown (DBAb) mice without 
the agent. J. Nat. Cancer Inst. 13: 505-531, 1952. 

(21) Andervont, H. B., and Dunn, T. B.: Attempt to detect a mammary tumor- 

agent in strain C mice by X-radiation. J. Nat. Cancer Inst. 10: 1157-1189, 

Hilberg, A. W.: Osteogenic sarcoma of mice. (Abstract.) Proc. Am. Assoc. 
Cancer Res. 1: 20, 1954. 

Cloudman, A. M.: Spontaneous neoplasms in mice. In Biology of the Labora- 
tory Mouse (Snell, G. D., ed.). Philadelphia, The Blakiston Co., pp. 168-233, 

(24) Slye M., Holmes, H. F., and Wells, H. G.: Tumors of the ovary in mice. 

J. Cancer Res. 5: 205-226, 1920. 

(25) Dunn, T. B.: Transplantable plasma cell neoplasm in strain C3H mice. (Ab- 

stract.) Proc. Am. Assoc. Cancer Res. 1: 13, 1954. 

(26) Stewart, H. L., Kaplan, H. S., and Bennett, J. G.: Report of two cases of 

identical primary tumors involving spinal nerve roots and meninges in strain 
NHO mice. J. Nat. Cancer Inst. 11: 177-197, 1950. 

(27) Cramer, W., and Horning, E. S.: On the association between brown degenera- 

tion of the adrenals and the incidence of mammary cancer in inbred strains of 
mice. Am. J. Cancer 37: 343-354, 1939. 

(28) Figge, F. H. J., Strong, L. C, Strong, L. C, Jr., and Shanbrom, A.: Fluores- 

cent porphyrins in harderian glands and susceptibility to spontaneous mam- 
mary carcinoma in mice. Cancer Res. 2: 335-342, 1942. 

(29) Davidsohn, I., and Stern, K.: Further studies on natural antisheep agglutinins 

in mice of inbred strains. Cancer Res. 10: 571-576, 1950. 

(30) Rawlinson, H. E., and Hankinson, H. W.: Stainable iron deposits in the 

epithelium of the mammary glands of mice. Anat. Rec. 102: 55-61, 1948. 

(81) Huseby, R. A., and Bittner, J. J.: A comparative morphological study of the 

mammary glands with reference to the known factors influencing the develop- 
ment of mammary carcinoma in mice. J. Cancer Res. 6: 240-255, 1946. 

(82) Deringer, M. K.: Spontaneous and induced tumors in haired and hairless 

strain HR mice. J. Nat. Cancer Inst. 12: 437-445, 1951. 
(88) Stewart, H. L. and Andervont, H. B.: Pathologic observations on the adeno- 
matous lesion of the stomach in mice of strain I. Arch. Path. 26: 1009-1022, 

Journal of the National Cancer Institute 


(34) Barnes, W. A., and Sisman, I. E.: Myeloid leukemia and non-malignant extra- 

medullary myelopoiesis in mice. Am. J. Cancer 37: 1-35, 1939. 

(35) Potter, J. S., Victor, J., and Ward, E. N.: Histological changes preceding 

spontaneous lymphatic leukemia in mice. Am. J. Path. 19: 239-253, 1943. 

(36) Budds, 0. C, Russell, E. S., and Abrams, G. E.: Effects of genetics and 

anesthesia upon granulocyte and agranulocyte levels in seven inbred mouse 
strains. Proc. Soc. Exper. Biol. & Med. 84: 176-178, 1953. 

Vol. 15, No. 3, December 1954 

586 proceedings: symposium on 25 years op 

Plate 42 

Figure 1. — Strain C58. Osteoid tissue surrounding original shaft of rib, which is 
more deeply stained. Osteoid encroaches on marrow cavity. Lesion developed in 
old mice on a synthetic diet. Hematoxylin and eosin. X 200 

Figure 2. — Strain C58, cyst of the pituitary, lying between pars intermedia and 
anterior lobe. A spontaneous lesion. Hematoxylin and eosin. X 50 



- ■ ■>' ■ 


316263—54 20 


588 proceedings: symposium on 25 years of 

Plate 43 

Figure 3. — A) Strain C3Hf, aged 14 days, untreated. All layers of retina are present. 
Optic nerve fibers emerging at right. Hematoxylin and eosin. X 300 

B) Strain C3Hf, aged 28 days, untreated. Rod layer and outer nuclear 
layer are missing. Optic nerve fibers are seen beneath retina. Hematoxylin and 
eosin. X 300 

C) Strain A, aged 1 year, exposed to 400 r X ray at birth. Rod layer 
and outer nuclear layer are missing. Optic nerve fibers are seen beneath retina. 
Hematoxylin and eosin. X 300 


***** m^ *5fe\** it ?f in 

*#>•*>***'" V ***e 

B * > 


Figure 3 



Dr. E. D. Murphy, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

It is particularly fitting that Dr. Thelma Dunn has been invited to speak in honor 
of Dr. Elizabeth Fekete and that she chose for her subject the problem of the mor- 
phologic differences between inbred strains, for these two investigators clearly share 
leadership in this field. In fact they are the pathologists whom pathologists consult 
on problems in this field. 

In most morphologic work it is a comfort to realize that if one goes back in the 
German literature one can usually find a classical morphologic study on almost any 
problem and one is saved a lot of work. The development of the inbred strains has 
been too recent for that, and it often seems to me that if Dr. Dunn or Dr. Fekete 
has not already carried out the basic studies, the work remains to be done. 

I was fascinated by the range of examples of morphologic differences between inbred 
strains that Dr. Dunn has shown us. It is apparent that cancer workers have appre- 
ciated and have used the inbred strains extensively in their work. It is also quite 
apparent that the inbred strains are and certainly will be of equal value in many 
other fields of investigation. Another field that stands out to me would be the group 
of degenerative diseases as contrasted, let us say, to infectious diseases about which 
we had an example in the first paper this afternoon. I was particularly impressed by 
the number of problems which morphology was able to delineate which demand further 
study. For example, the kidney lesions induced by urethan offer a possible experi- 
mental analogue to glomerular lesions in man. 

The term classical, as applied to these recent studies, refers not to the use of tradi- 
tional histologic techniques, but to the degree of thoroughness and meticulousness. 
It is quite apparent that modern classical studies employ the techniques required to 
answer the questions put, including the histochemical methods which are undergoing 
such rapid development at present. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Session II. Mammalian Genetics; 1929 
as Viewed from 1954 

Chairman, Dr. Clarence C. Little, Director, 
Roscoe B. Jackson Memorial Laboratory, Bar 
Harbor, Maine 

Speaker: Dr. W. E. Castle 

The Part of Mammalian Genetics in Founding The Jackson Memorial 


Journal of the National Cancer Institute. Vol. 15, No. 3, December 1954 

Introduction to Dr. Castle 
Dr. C. C. Little 

It is not often that a man who had an important part in the birth of a new science 
is still contributing actively to its progress almost 55 years later. 

After the rediscovery of Mendel's Law, Dr. Castle was one of the first to apply 
its principles to mammalian genetics. Over a longer period than anyone else he has 
continued research in that field with fruitful and stimulating results. 

In this room are many of his academic "children," "grandchildren," and "great- 
grandchildren," who have in their lives reflected the impelling stimulus and the patient 
effort for which he was the example. 

In many ways the Jackson Laboratory represents a happy and vigorous attempt to 
realize the opportunities which had their ancestry in the early days of his teaching. 

From the basement of old Lawrence Hall at Harvard to the Bussey Institution and 
out into the world of experimental biology the impetus given to us by this great and 
humble teacher has made its way with faith and strength. 

His being here today is, for me personally, the fulfillment of a dream of long stand- 
ing. It is entirely proper that this celebration of our 25th birthday should be graced 
by his presence. I hope that both by word and by inner awareness, he will know 
how genuine and how lasting is our respect and affection for him as a man, and as a 
catalyst expressed through his students. 


The Part of Mammalian Genetics in 
Founding The Jackson Memorial Lab- 
oratory * 

W. E. Castle, University of California, Berkeley, 

I have been reading recently a little book about the life and work of 
Sir Alexander Fleming, the discoverer of Penicillin. He is therein repre- 
sented as holding views concerning the method by which science advances, 
which I find interesting and worthy of further consideration. 

He thinks the individual is all-important in the origination of new 
ideas in science. These ideas result from long-continued thought and 
experimentation on the part of a gifted and well-trained mind, on a subject 
with which past experience has made it well acquainted. 

Fleming's own experience illustrates this. He had spent long years in 
the study of immunity, natural and acquired, to bacterial infections. 
When by accident a spore of the mold Penicillium notatum got into one of 
his cultures and promptly checked growth of the bacteria therein, he 
recognized at once that this mold was capable of becoming an important 
agency in arresting bacterial infection. 

Here was the germ of a great idea but it took 10 years of work on his 
part, and that of biochemists, to fully evaluate the idea and bring it to 

The origination of a revolutionary new idea and its evaluation are quite 
distinct processes. For the first, a great mind alone suffices. For the 
second process minds of lesser capacity may be adequate, if the means 
employed are characterized by unbiased integrity and sound judgment. 

Let us turn now from this generalization to the consideration of a 
specific case, the status of experimental biology in the year 1900. Four 
men of genius, in the preceding half century, had formulated ideas of 
revolutionary importance in biological thinking. These men were 
Darwin, Weismann, Mendel, and de Vries. 

Darwin's new idea was the origin of species by natural selection acting 
on spontaneous variations. 

Weismann's new idea was the concept of germ-plasm distinct from 
body-plasm, the germ-plasm alone being continuous and immortal, 
whereas the body-plasm was destined to die when the life of the individual 

« Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 27, X954. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 


came to an end. Weismann conceived the continuity of life to depend 
upon heredity, the basis of which was in the chromosomes of the nuclei 
of the germ cells. 

Mendel's new idea antedated Weismann's, but it had to lie dormant 
for a quarter century while the ideas of Darwin and Weismann were 
conditioning the intellectual soil for its germination and fruition. 

Mendel's idea, which his careful experiments with garden peas demon- 
strated to be correct, was that of the mutual independence of inherited 
characters in transmission. He did not attempt to localize in the germ 
cells the material basis of inheritance, as Weismann later did. For this, 
in Mendel's time, knowledge of cytology was inadequate, but the lack 
had been made good when Weismann formulated his ideas. 

De Vries' new idea was that of mutation; that new hereditary characters 
come into existence abruptly, not gradually as most Darwinians thought, 
and that thereafter they remain constant. 

These four basic concepts, put forward by minds of great originality, 
each working largely alone, awaited evaluation by biologists in the year 

At that time it was my good fortune to be alive and just getting a 
foothold in biological investigations, which were to qualify me as a junior 
member of the evaluating team. Senior members of the team, who 
already had major accomplishments to their credit, were: E. B. Wilson, 
author of The Cell, and T. H. Morgan, experimental zoologist, at Columbia 
University; E. G. Conklin, embryologist at Princeton University; and 
Ross Harrison, active at Yale in tissue culture and transplantation. These 
men and a few others founded the Journal of Experimental Zoology for 
the publication of their findings. 

All these experimenters, except Harrison and me, have now completed 
their work and passed off the stage. Harrison and I, fortunately for us, 
are still viewing the scene from the wings. 

The first of these great ideas to attain evaluation, nearly unanimous in 
its favor, was the evolution theory of Darwin. This favorable verdict was 
expressed in a symposium in 1909 on Fifty Years of Darwin. It was my 
privilege to have a part in this symposium. It was my further privilege 
in 1950 to be still alive and to join in the celebration of Fifty Years of 
Validation of Mendel's great idea. Today I rejoice in being present at a 
celebration of the founding of this center of research which in its 25-year 
history has contributed much to the validation of Mendel's principles. 

The second basic idea, Weismann's germ-plasm theory, received 
substantial support in an ovarian transplantation experiment by John C. 
Phillips and me, reported in 1909. We showed that an ovary taken from 
an immature black guinea pig and transplanted into the body of an 
albino, retained there its distinctive character — the mature ova producing 
black offspring even when the male parent as well as the foster mother 
were both albinos. 

Journal of the National Canrer Institute 


The third basic idea (that of Mendel) of the independent inheritance 
of unit characters, was the prime object of investigation in my laboratory 
in the early years of the century. 

My pupils and I verified it first in the case of albinism in mice, later in 
numerous other coat characters in mice, rats, guinea pigs, and rabbits. 
Parallel investigations were at this time being made in England, by pupils 
of Bateson, that furnished additional evidence of the validity of Mendelian 
principles. Indeed it would be more correct to say that our work in 
America complemented and extended that of the pupils of Bateson. The 
initiative was taken by them. 

The fourth basic idea (that of de Vries) of mutation as a source of new 
and stable variations received strong support when Muller, a pupil of 
Morgan's, showed that mutations could be produced artificially by 
subjecting germ cells to bombardment with X rays. 

What, then, we may inquire was the status of Mammalian Genetics in 
1929, when the Jackson Memorial Laboratory was founded? 

1) It had been shown that the inheritance of the coat characters of 
rodents and other mammals depended upon the transmission in the germ 
cells of material bodies which had come to be called genes or inheritance 

A gene might by mutation suddenly take on a new form. The new 
form (termed an allele) would now conform with Mendel's law in relation 
to the antecedent form, being either dominant over it or, more often, 
recessive to it. New combinations of genes might arise, or be produced 
at will, as a result of mutations in two or more different genes. When 
the genes involved in mutation were located in different chromosomes, 
the combinations which emerged from crosses would conform in their fre- 
quencies with the laws of chance, as in Mendel's experiments with peas. 
But when two genes were located in the same chromosome, they would 
have a tendency to stay together in transmission, a phenomenon called 
linkage, not encountered by Mendel, but first observed by Bateson and 
Punnett in sweetpeas, and later both observed and correctly interpreted 
by Morgan and his pupils in Drosophila. 

In mammals, linkage was first observed independently by Haldane and 
myself in mice. Later, cases of linkage were found in my laboratory in 
rats and rabbits also. Since 1929 the study of linkage in rats has been 
one of my major interests. Five linkage groups have now been identified. 
Meanwhile, in the Jackson Laboratory the number of linkage groups in 
mice has increased to 10, as new mutations have been discovered. Dr. 
George Snell deserves credit for most of these discoveries of linkage in 
mice. He did the initial work in this field at the Bussey Institution, 
incorporating the results in his doctor's thesis. 

2) The genetic interpretation of intermediate or blending inheritance 
was long a baffling problem, but by 1929 it had been resolved, largely on 
the basis of experiments with plants, into a case of multiple genes acting 
simultaneously but without demonstrable dominance of one allele over 

Vol. 15, No. 3, December 1954 

598 proceedings: symposium on 25 years op 

An intensive study had been made of size inheritance in rabbits, a 
typical blending character, by crosses between races of large and small 
body size respectively. Attempts to locate a gene affecting body size in 
four different chromosomes had proved fruitless. However P. W. 
Gregory and I were able to show from an embryological study that genes 
located in the chromosomes must be assumed to be involved in deter- 
mination of body size in rabbits, since size difference involves a difference 
in rate of development of the fertilized eggs, and this difference is influenced 
by sperm as well as egg. Large race egg plus small race sperm has sub- 
stantially the same rate of development as small race egg plus large race 
sperm. In both cases intermediate body size results. Further discoveries 
in this field followed the founding of the Jackson Laboratory. 

3) It was known in 1929 that close inbreeding of mammals does not of 
necessity involve racial deterioration, as had been supposed earlier to be 
true. This conclusion had been reached in the case of Drosophila by the 
50-generation experiment of brother X sister mating made in my labora- 
tory, the first breeding experiment in which this now famous fly was 
employed. Later, Dr. Helen Dean King verified this finding for mammals 
by inbreeding rats for a like extended series of generations. Theoretically 
such prolonged inbreeding should render a race completely homozygous 
and thus very uniform in its physiological behavior. 

This was the cornerstone on which in 1929 Dr. Little built the program 
for cancer research at the Jackson Laboratory. He had already created 
a long inbred strain of mice, the now famous d br non-agouti race (DBA), 
which he began inbreeding at the Bussey Institution in 1909, and which 
is still going strong. He then undertook to produce other inbred strains 
of different but uniform genetic constitution. The resulting races are now 
being used because of their known uniform genetic constitution and 
consequent uniform physiological behavior in genetic and cancer investi- 
gations throughout the world. 

With this equipment of generalized ideas about mammalian genetics, 
and with his personal experience in cancer investigations in mice, Dr. 
Little was ready and eager in 1929 to resume the investigations which had 
been interrupted to some extent by his two college presidencies. Only 
a laboratory with an assurance of adequate financial support was needed, 
and this fortunately was forthcoming — though not spontaneously, but 
only as a result of the enthusiastic personal leadership, which has always 
characterized Dr. Little. 

I have known Dr. Little for a long time, longer perhaps than anyone 
else present. We first came into contact when as an undergraduate in 
Harvard College he enrolled in my course in genetics. I set him to 
studying color inheritance in mice and this later became the subject of 
his doctor's thesis. I was impressed with his energy, enthusiasm and 
resourcefulness, and the next year made him my assistant in giving the 
course. He was then captain of the track team, his own feat of strength 
being, I believe, putting the shot farther than anyone else. His enthu- 
siasm for genetics was communicated to fellow members of the track 

Journal of the National Cancer Institute 



team, with the result that enrollment in my course increased considerably. 
This was a direct evidence of his gift for leadership. Further evidence of 
it followed to my benefit. 

He became secretary to President Lowell, on the retirement of President 
Charles W. Eliot, in whose administration I had been advanced to a 
Professorship of Zoology. Little must have whispered in the president's 
ear that genetics was hot stuff, for I was invited to give a course of lectures 
on the subject at the Lowell Institute in 1910 and these lectures formed 
the basis of my first publication in book form, a little book entitled 
Heredity in Relation to Evolution and Animal Breeding, New York, D. 
Appleton & Co., 1913. 

Little's mice, with the rest of our animal colony, were transferred in 
1909 to the Bussey Institution at Forest Hills, alongside of the Arnold 
Arboretum. There a fruitful association was formed between my pupils 
and those of a brilliant new colleague, the late and lamented E. M. East, 
Professor of Plant Genetics. 

Little and I were joint authors of several short publications which were 
outgrowths of his studies of mice, and in 1913 his full report on the subject 
was published (1) with excellent colored plates in Publication No. 179 
of the Carnegie Institution of Washington, which was assisting my work 
as a Research Associate. 

Little's energy and enthusiasm led him into several minor projects, 
which I encouraged him to undertake, although at the time we lacked 
adequate facilities for their completion. These projects were: crosses of 
dogs with different racial conformation and instincts; crosses of different 
races of domestic pigeons ; and investigation of the sterility of the exception- 
ally produced male tortoise-shell cat. 

The pigeon project, which we were unable to carry out, was later taken 
up at the University of Wisconsin with notable success by L. J. Cole and 
M. R. Irwin. 

Little's fondness for dogs and eagerness to experiment with them 
persisted unrealized until recent years when happily facilities for such 
work were created at the Jackson Laboratory. We shall hear more on 
this subject from Dr. Scott. 

Meanwhile, Dr. Tyzzer at Harvard Medical School had been experi- 
menting with crosses of Japanese waltzing mice involving susceptibility 
to transplantation of a tumor peculiar to that race of mice. He appealed 
to us at the Bussey for help on a genetic problem that was baffling him. 
I recommended Dr. Little to him as the person best qualified to render 
such assistance, and this Little did effectively. So this is how he became 
involved in the study of cancer in mice — an involvement which still 

In 1917, the United States entered World War I and both Professor 
East and Dr. Little were transferred to Washington in government 
service. When the war was over, Little became attached for a time to 
the staff of C. B. Davenport, head of the department of genetics of the 
Carnegie Institution at Cold Spring Harbor, New York. But he presently 

Vol. 15, No. 3, December 1954 


600 proceedings: symposium on 25 years of 

moved on to the presidency of the University of Maine, and later to that 
of the University of Michigan. 

But he was, I think, homesick for research in an institution like the 
Bussey of his earlier days. And since time on the Bussey clock was 
running out, there was nothing to do but found such an institution him- 
self. This he proceeded to do, leaving college presidencies to those more 
interested in academic matters than in biological research. 

In the founding and fostering of the Jackson Memorial Laboratory, Dr. 
Little has spent a busy quarter century, with results which we shall learn 
more about in the next 2 days. We wish for him, and the Jackson Labo- 
ratory, continued success in a project of major humanitarian interest. 

Let us now turn from a consideration of the state of mammalian genetics 
prior to 1929 to a brief discussion of its subsequent development. No 
revolutionary new discoveries in mammalian genetics have been made in 
this period, but progress has been made in extending and clarifying 
knowledge in fields already explored (2). 

A greatly increased number of mutated genes is now known in mam- 
mals in which mutations had been previously reported, as for example in 
mice, in regard to which animal work has been centered in the Jackson 

Mutations have also been reported in many additional species; for ex- 
ample, the mink, which is now being reared in captivity on a large scale 
for its valuable fur. Thousands of mink are now carefully inspected each 
year by breeders and furriers anxious to discover any new or improved 
qualities of the fur. This has resulted in finding many mutations that 
might otherwise have been overlooked. 

The economic value of some of these mutations has led to the formation 
of a Mutation Mink Breeders' Association. An excellent little book on 
Genetics of the Ranch Mink has been written by Dr. R. M. Shakelford of 
the University of Wisconsin, a pupil of the late lamented Professor L. J. 
Cole. Shakelford finds the chromosome number of the mink to be 14, 
much lower than that of laboratory rodents, which is upward of 20. He 
reports on 14 different mutations, with one case of linkage. 

A word in passing might be said about Professor Cole, who had a large 
share in the advancement of mammalian and avian genetics in his lifetime 
through teaching and research of distinguished character. While studying 
at Harvard he majored in a zoological field which did not directly involve 
genetics, but became interested in the newly developing field of genetics, 
and later gave to it his entire attention. 

A similar interest in genetics was aroused in the late Professor H. E. 
Walter of Brown University, who (though majoring in a different zoological 
field) took a course with me and did experimental work in genetics while 
studying in the Graduate School in Cambridge. Later he became eminent 
as a teacher of genetics and author of a widely used textbook of genetics. 

Returning from this digression, let us inquire: What are the more im- 
portant advances in mammalian genetics made since 1929? On the 
theoretical side, these relate among other things to size inheritance, 

Journal of the National Cancer Institute 



heterosis and statistics. On the practical side, they include breeding plans 
for the more efficient production of domestic animals of the types desired. 

Size Inheritance 

I have already discussed the state of knowledge of size inheritance prior 
to 1929. It was known that size genes existed but attempts to identify 
or localize them in linkage systems had proved fruitless. 

An important discovery was now made by C. V. Green (3), a charter 
member of the staff of the Jackson Laboratory, whose brilliant work was 
brought to an untimely end by his tragic death in 1935. 

Green had already begun at Ann Arbor, under Dr. Little's direction, 
the study of a cross between a species of small mouse from China, Mus 
bactrianus, and Little's much larger dilute-brown inbred strain of Mus 
musculus. The two parental races differed not only in size but also in 
three color genes, A, B and D, of which the small race carried dominant 
alleles and the large race recessive alleles. Green showed that, in F 2 and 
backcross populations, individuals which were homozygous for the reces- 
sive alleles, b for brown, and d for blue dilution, were larger than those 
carrying the corresponding dominants. He assumed that this was due to 
linkage between a gene, or genes, for large size and color genes b and d. 

Since Green had found in a mouse cross what I had previously looked 
for in vain in rabbit-size crosses, I now undertook to repeat Green's ex- 
periment using the same dilute-brown race used by Green as the large 
size parent, stock of which was kindly supplied by Dr. Little. For the 
small race parent, I used, not the captive wild race of Mus bactrianus, 
but one of its supposed domestic derivatives, the Japanese waltzing mouse, 
of which W. H. Gates was making an intensive study in my laboratory. 
This investigation was shared by my pupils at the Bussey Institution — 
W. H. Gates, S. C. Reed and L. W. Law. 

We confirmed Green's findings as to fact, but gave them a different 
interpretation (4). Our interpretation was that color genes, when they 
are associated with size differences, owe the association, not to linkage 
with specific genes for size, but to direct physiological action of their own. 
This interpretation was verified in further studies of size crosses made by 
me in Berkeley, after my retirement and years after Green's death, again 
largely with mouse stocks kindly supplied for the purpose by the Jackson 

I used in these studies three of the same criteria of size used by Green; 
namely, body weight, body length, and tail length. I found that some 
color genes are without any discoverable influence on body size, others 
increase it, and still others decrease it. Gene mutations with no dis- 
coverable influence on body size are non-agouti and albino; those which 
increase body size by all three criteria are brown, blue dilution, and 
yellow, the last being a dominant mutation, effective only when heter- 
ozygous as is well known (the homozygote being lethal). Mutations 

Vol. 15, No. 3, December 1954 

602 proceedings: symposium on 25 years op 

which decrease body size by all three criteria are pink-eye, pink-eye 2 , 
leaden, short ear, and dwarf, all recessives. 

Another finding of Green's (5) which I was able later to verify was this. 
Certain genes that have an influence on general body size show greater 
influence on some body regions than on others. They thus show a 
specific local influence on growth, as well as a lesser general influence. 
This is well illustrated in the case of the mutation d, blue dilution, which 
Green in his original experiment showed to increase general body size, 
but to increase tail length out of all proportion to body weight and body 
length. This finding of Green's I was able fully to confirm. The muta- 
tion short ear, as its name indicates, has a specific local influence in 
reducing ear size, greater than its effect on body weight and body length, 
as had been shown by Snell. 

Both Green and I, in our studies of size inheritance in mice, were in- 
fluenced by the conclusions reached by Wright in a statistical analysis of 
my data on size crosses in rabbits (6). Wright recognized the existence 
of general, group, and special size factors. What we had shown was 
that general growth factors (such as mutation d) may also function as 
special growth factors in regard to tail length or cranial dimensions 
(Green), or ear length (Snell). 

Wright's conclusions are unquestionably still valid with regard to the 
categories of genetic influence in size being mostly general, but to some 
extent exerted regionally on groups of parts, or on individual special parts. 
However, it does not follow of necessity that these different kinds of 
influence are exerted by different genes. 

To find out whether color genes influence size in mammals other than 
mice, I extended my studies in Berkeley to rats and rabbits also. 

The color mutation, brown, which in mice had been found to have 
greater influence in increasing body size than any other investigated, was 
found to be a gene for increased body size in rabbits and in rats also. I 
am curious to know whether this is true also in dogs, but have had no 
opportunity to investigate it. Perhaps you at the Jackson Laboratory 
will have. 

Heterozygous Phenotypes 

In animal breeding a striking difference in phenotype (appearance) be- 
tween a heterozygote and a homozygote for a particular color gene has 
been well known ever since W. Bateson described the genetics of the Blue 
Andalusian fowl. In such a case, if breeders happen to prefer the heter- 
ozygote, they are dealing with an unfixable variety, one which will breed 
true only to the extent of half of the progeny, which in turn will breed 
in the same way. I discovered man} 7 years ago, that this is true in the 
English rabbit, where the fanciers' ideal is an animal with a maximum 
number of small spots well distributed over the body, as in a Dalmatian 
(coach) dog. 

When standard English rabbits are interbred, they produce only about 
50 percent of standard English young, together with 25 percent of un- 

Journal of the National Cancer Institute 



spotted (self) young, and 25 percent of animals with too few spots to 
conform with the English standard. These last are known as Charlie 
Chaplins, being regarded as a joke, so far as conformation with the 
standard for English spotting is concerned. Both the Charlie Chaplins 
and the selfs are homozygotes and, if crossed with each other, will produce 
100 percent of heterozygotes of the standard English type. 

But this short-cut in mass production of standard English is not recom- 
mended to rabbit fanciers who desire to produce show specimens, since 
fine points in the standard pattern, due to modifying genes, may be lost 
in the process of "sight unseen" mass production. 

I have found a parallel case to that of the English rabbit in the Palomino 
horse, since my migration to California where Palominos are very popular. 
The breeder's ideal Palomino has a golden-yellow body color with white 
mane and tail. The actual Palomino which most nearly approximates 
this ideal is heterozygous for a dominant-dilution gene (D) on a very 
specific background of other color genes. A horse homozygous for D is 
too dilute to be desirable. It is called an albino. In addition to having 
a very pale, ivory-white coat, it has a pink skin and blue eyes. Palominos 
interbred produce 50 percent of Palominos, and 25 percent each of albinos 
and horses of full color. Both of these latter classes are homozygous as 
regards gene D — one having it, the other lacking it. Breeding them 
together produces 100 percent of heterozygotes, the Palomino genotype. 
But again, as in the case of the English rabbit, this practice is not to be 
recommended to fanciers who want to produce show specimens, since such 
can be produced only on a specific background of color genes. 

The genetic background on which the dilution gene acts in the produc- 
tion of the preferred type of Palomino, is identical with that of a sorrel 
horse having a light mane and tail. It is homozygous for the recessive 
brown gene (6) which distinguishes chestnut and liver from bay and black. 
The distribution of its brown pigment is restricted by two other genes, 
the dominant ancestral-pattern gene (A) which makes the difference 
between bay and black, and the recessive restriction gene (e) which, when 
homozygous, replaces black-brown pigment with red-yellow pigment in 
the coat generally. The dilution gene (D) when heterozygous reduces the 
intensity of the red body-color of a sorrel to the golden yellow of a 
Palomino ; when homozygous it reduces the red body-color too much to 
the near white of the so-called albino. Its effect, when heterozygous, on 
the light mane of a sorrel is to render it still lighter, approximating the 
"white" of the Palomino standard. 

The net results of the interaction of color genes in this heterozygous 
genotype is a horse with a diluted red coat, described by the fancier as 
"golden yellow, the color of newly minted gold," and an almost "white 
mane and tail." 

The preference of cattle breeders for a type of conformation which 
(without their knowing it) belongs to a heterozygous genotype, may lead 
to disastrous consequences. This has been shown by Gregory et at. (7) in 
the case of cattle which are heterozygous for a dwarf gene. Homozygous 

Vol. 15. No. 3, December 1954 
316263—54 21 

604 proceedings: symposium on 25 years of 

dwarfs are either lethal and so a total loss, or of greatly reduced value, yet 
they constitute one quarter of the calves produced by heterozygous 
parents of the preferred phenotype. The detection of heterozygotes for 
dwarfness among herd sires thus becomes important to cattle breeders, 
who may then take measures to avoid mating a heterozygous bull to 
heterozygous cows. 


Important advances have been made since 1929 in regard to heterosis 
(hybrid vigor). The basic discoveries in this field have been made in the 
study of plants or animals such as Drosophila and Habrobracon. But 
these have found important applications in animal breeding. 

Everyone is familiar with the case of hybrid corn, an astonishing ex- 
ample of the improvement in an important field crop to be derived from 
prolonged inbreeding followed by a single generation of cross breeding. 

Breeders of pigs, following the lead of the corn breeders who supply 
them with more abundant feed for their pigs, have secured similar though 
less striking benefits from crossing purebred breeds of swine. 

In the early days of the study in maize of the effects of inbreeding and 
cross breeding made by Shull and East, it was thought that the increased 
vigor of hybrids was due exclusively to the physiological superiority of 
heterozygotes over homozygotes. Hence the term heterosis was coined as 
a short name for heterozygosis. 

But further studies (£), particularly those of East and his pupils, have 
given to heterosis a more extended significance. It is now thought to in- 
volve four different agencies. 

1) Dominance of favorable genes over their unfavorable recessive 
alleles. This is ordinary dominance. In this case, a homozygote has the 
same physiological action as a heterozygote. 

2) Over-dominance of favorable genes, which renders their action 
greater as heterozygotes than as homozygotes. This is the original 
conception of hybrid vigor. 

3) Linkage of dominant favorable genes with recessive unfavorable 
genes. Under prolonged inbreeding both sorts will become fully operative 
and the net result will be deterioration to the extent that unfavorable 
genetic action surpasses favorable genetic action. Crossing of two differ- 
ent inbred lines will suppress the action of all unfavorable recessive genes 
carried by both parents and allow unhampered action of all favorable 
genes whether ordinary dominants or over-dominants. A maximum of 
hybrid vigor will result. 

4) Gene interaction. In some cases favorable action of a gene may be 
increased or diminished by interaction with a different gene, particularly 
with one located in a different chromosome. 

Opinions differ as to the relative importance of these four agencies. 
It is probable that linked recessive unfavorable genes are of commoner 
occurrence in maize than in mammals, and their suppression of major 
importance in hybrid corn but not in cross-bred swine, in which over- 
dominance is more probably of major importance. 

Journal of the National Cancer Institute 




In the field of study of linkage in mammals, in which I am currently 
much interested, an important advance is sure to result from a finding by 
Kramer et al. (9) in barley that two linkage groups, seemingly independent, 
may exist at opposite ends of the same chromosome, if a sufficiently long 
portion of the chromosome devoid of identified genetic loci intervenes 
between them. 

That this principle is valid for mammals is shown by an unpublished 
investigation of the linkage interrelations of three genes in Chromosome IV 
of the rat. The genes in question are wobbly, hairless, and naked. The 
gene hairless lies between the other two, about equally distant from each, 
and is clearly linked with both. If a test for linkage between the end 
genes had been made, without knowledge of the existence of the inter- 
mediate gene hairless, the erroneous conclusion would have been reached 
that these genes were located in different chromosomes, since they give no 
indication of linkage with each other. 

Consequently many earlier linkage tests that gave negative results must 
be regarded^as inconclusive and will merit re-examination as new genes 
are discovered and their linkage relations studied. 


I shall not undertake the discussion of advances made in the last quarter 
century in the field of statistics, though these are of great importance in 
the field of mammalian genetics in general, of human genetics in particular, 
and in discussions of evolution — as well as in practical plans for the 
breeding of plants and animals. 

Masters in these several fields are with us here and we shall await the 
statement of their findings in the programs to follow. 


(1) Little, C. C: Experimental studies of the inheritance of color in mice. Publ. 

No. 179, Wash., D. C, Carnegie Institution of Washington, 1913. 
{2) Dunn, L. C: Genetics in the 20th Century. New York, The Macmillan Co., 

1951, 634 pp. 
(3) Green, C. V.: Linkage in size inheritance. Am. Nat. 65: 502-511, 1931. 
U) Castle, W. E.: Size inheritance. Am. Nat. 75: 488-498, 1941. 
(5) Green, C. V.: Further evidence of linkage in size inheritance. Am. Nat. 67: 

377-380, 1933. 
{6) Wright, S.: General, group, and special size factors. Genetics 17: 603-619, 

(7) Gregory, P. W., Roubicek, C. B., Carroll, F. D., Stratton, P. O., and 

Hilston, N. W.: Inheritance of bovine dwarfism and the detection of heterozy- 

gotes. Hilgardia 22: 407-450, 1953. 
(5) Gowen, J. W.: Heterosis. Ames, Iowa, Iowa State College Press, 1952, 552 pp. 
(9) Kramer, H. H., Veyl, R., and Hanson, W. D.: The association of two genetic 

linkage groups in barley with one chromosome. Genetics 39: 159-168, 1954. 

Vol. 15, No. S, 



proceedings: symposium 


Dr. C. C. Little 

Dr. Castle was too modest to tell you one of his most brilliant pieces of work which 
was early in the game, I think in 1903, when he recognized that the approximate 
equality of the sexes was probably due to the backcross type of Mendelian one to one 
ratio cross, thereby placing the theory of sex determination on the basis of Mendelism. 
The fact was later confirmed by McClure and others by cytological investigations. 
Dr. Castle said he was "in the wings." Well, the "wings" that he has have many 
strong feathers and those of us who have watched his flight in science will do so I am 
sure for a great many more years to come with very deep affection and very great 
admiration. It is the flight of a very kind and self-effacing eagle. 

Session III. Genetic Control of Devel- 
opmental Patterns 

Chairman, Dr. W. H. Gates, Professor Zoology, 
Entomology, and Physiology, Louisiana State Uni- 
versity, Baton Rouge, La. 

Speaker: Dr. Earl L. Green 

Quantitative Genetics of Skeletal Variations in the Mouse. I. Crosses 

Between Three Short-Ear Strains (P, NB, SEC/2) 
Discusser: Dr. Paul B. Sawin 

Speaker: Dr. Salome Gluecksohn-Waelsch 

Genetic Control of Embryonic Growth and Differentiation 
Discusser: Dr. Morris Smithberg 

Speaker: Dr. Meredith N. Runner 

Inheritance of Susceptibility to Congenital Deformity — Embryonic Insta- 
Discusser: Dr. Donald W. Bailey 


Quantitative Genetics of Skeletal Vari- 
ations in the Mouse. I. Crosses Be- 
tween Three Short-Ear Strains (P, NB, 

SEC/2) *• 2 

Earl L. Green, Department of Zoology, Ohio State 
University, Columbus, Ohio, and Division of Biology 
and Medicine, U. S. Atomic Energy Commission, 
Washington, D. C. 

Laboratory strains of the house mouse are rich in variations of skeletal 
components, both within and between strains. The variable sites are 
distributed over the appendages, the pelvic and pectoral girdles, and the 
axial skeleton from the nose to the tip of the tail. Some sites present 
a discrete presence and absence variable, such as the presence or absence 
of the omosternal bones; some present a continuous measurable variable, 
such as the length of the nasal bones ; and some present a discrete numeri- 
cal variable, such as the number of digits or the number of vertebrae. 

Variations in the skeleton of vertebrates have been the subject of 
numerous investigations. Anatomists, embryologists, and geneticists 
have recorded and studied the variations in the skeleton of horses, sheep, 
pigs, dogs, and of various primates, including man. It is fitting to note, 
during this 25th year of the Jackson Memorial Laboratory, that my own 
interest in the genetic and nongenetic causes of variation in the skeleton 
of the mouse started at the Jackson Laboratory. In the summer of 1938, 
following the lead of Dr. Paul B. Sawin, who had discovered interesting 
variations in the skeleton of the rabbit, I visited the Jackson Laboratory 
to survey the available inbred strains of mice for skeletal variations. I 
found that there were marked differences between inbred strains as well 
as considerable variability within strains with respect to the composition 
of the axial skeleton. Some strains, such as C57BL, were found to have 
13 thoracic and 6 lumbar vertebrae. Some other strains, such as DBA 
and C3H, were characterized by 13 thoracic and 5 lumbar vertebrae. 
One strain, P, had 12 thoracic and 6 lumbar vertebrae; and one strain, 
BALB/c, had 13 or 14 thoracic and 6 or 5 lumbar vertebrae. No strain 
studied then or since has been absolutely uniform with respect to the 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 28, 1954. 

* This investigation was supported in part from funds granted to the Ohio State University by the Research 
Foundation for aid in fundamental research, and in part by research grants (Nos. RG-2993 and A-544) from the 
National Institutes of Health, U. S. Public Health Service. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 19S4 

610 proceedings: symposium on 25 years of 

composition of the axial skeleton. Usually, 5 to 10 percent of the mice 
in each strain have turned out to depart from the principal type. 


The multiplicity of forms in the skeleton or any other organ system 
attracts our attention for several reasons. 

First, what is the genetic basis for the differences between strains? 
Do the differences "mendelize" in any familiar pattern? Are the differ- 
ences associated with any named genes? 

Second, is the variability within inbred strains exclusively nongenetic? 
What is the relative importance of genetic and nongenetic causes of varia- 
bility within strains? Can the nongenetic causes of variability be 

Third, in what ways are the developmental patterns different for mice 
of different skeletal types? How do the differences arise? 

Fourth, what experimental agents, physical or physiological, applied to 
eggs or to developing embryos can alter the course of development 
sufficiently to produce mice with different skeletal types? 

Fifth, what are the evolutionary implications of the extensive poly- 
morphism in the skeleton of laboratory strains of mice? 

Our present knowledge is too meager to answer any of these questions 
with assurance. Partial answers to some of the questions have been 
provided by several previous studies. The numerical variations at the 
lumbosacral border in strain BALB/c mice have been studied by Green 
(1) and in the C57BL strain by Searle (2). In both strains, nongenetic 
causes greatly outweigh genetic causes of variation. Crosses between 
strains have shown that both genetic and nongenetic causes are concerned 
in variation at the lumbosacral border. In a cross between C57BL and 
BALB/c, it was found that 3 or more blocks of genes are required to 
account for the difference between the strains [Green (3)]. In a cross 
between C3H and C57BL, a significant difference between reciprocal 
hybrids appeared, with the hybrid offspring tending to resemble the 
strain of the female parent [Green and Russell (4)]. 

An attempt to identify the nongenetic causes of variation at the lumbo- 
sacral border in the BALB/c strain yielded negative results for all forces, 
among those tested, acting on litters as units except for age of mother, 
and for all forces tested acting on horn-mates as units. It appeared that 
whatever the agents may be they act upon individuals as units [Green 
and Green (5)]. 

Named genes in the mouse have been discovered to affect the numerical 
variation in parts of the axial skeleton. These include, among several 
others: anury [Dobrovolskaia-Zavadskaia (6); Chesley and Dunn (7)], 
Danforth's short tail [Gluecksohn-Schoenheimer (8)], stub [Dunn and 
Gluecksohn-Schoenheimer (9)], screw tail [MacDowell et al. (10)], short 
ear [Green and Green (11)], luxate [Carter (12)], and luxoid [M. C. Green 

Journal of the National Cancer Institute 



The purpose of this paper is to describe some breeding experiments 
aimed at answering the first question — the genetic basis of differences 
between inbred strains of mice. One may approach this question in 
either of two ways. The variable sites may be studied singly or in com- 
bination. Given that the time and effort to be devoted to the study are 
in some way limited, one may study a large number of variable sites in 
the mice of the parental, hybrid, and later segregating generations at the 
expense of being unable to study large numbers of mice. Or one may 
study a large number of mice at the expense of reducing the number of 
variable sites to one, two, or three. For the purpose of this discussion, 
one site only, the position of the sacrum, will be referred to. 

The position of the sacrum may be most easily defined by the number 
of vertebrae which precede the first sacral vertebra. The number of 
presacral vertebrae — cervical, thoracic, and lumbar — in the mouse may be 

25, 26 or 27. Numbers greater or smaller than those are extremely rare. 
Some vertebrae are asymmetrical so that they resemble sacral vertebrae 
on one side, lumbar vertebrae on the other. The position of the sacrum 
may still be denoted by the number of presacral vertebrae as 25/26 or 
26/25, or as 26/27 or 27/26, where the numbers are given as Right/Left. 
When the 26th vertebra is asymmetrical, as in the first two cases, the 
class of mice of this sort may be denoted as A x ; when the 27th vertebra 
is asymmetrical, as in the latter two cases, the class of mice may be 
denoted as A 2 . In a few mice in crosses between strains, symmetrical 
vertebrae occur at the lumbosacral border which are neither clearly 
lumbar nor clearly sacral types. These vertebrae have transverse proc- 
esses which do not point forward and downward like those of lumbar 
vertebrae, and which do not extend sidewise to articulate with the in- 
nominate bones like those of sacral vertebrae. These few mice have 
been classed with the Ai or A 2 types, on the ground that Ai and A 2 repre- 
sent intermediate, as well as asymmetrical, classes. With this notation 
there are five sacral positions with 25, A x , 26, A 2 , and 27 presacral vertebrae. 

Differences Between Strains in Position of Sacrum 

Seven inbred strains have been recently sampled. Each strain exhibits 
some variation in the position of the sacrum. In no two of the strains is 
the variation identical. The proportions of mice with 25, Ai, 26, A 2 , and 
27 presacral vertebrae change from strain to strain (table 1). There are 
some similarities, however. Three of the strains (P, C3H, DBA) have 
high percentages of mice with 25 presacral vertebrae; two of the 
strains (C57BL, NB) have high percentages of mice with 26 presacral 
vertebrae. The other two strains have sizable percentages of mice with 

26, A 2 , and 27 presacral vertebrae. 

Some sources of variability, both in inbred strains and in crossbred 
generations, may be detected at once when tables are prepared with more 
detail than table 1. Among these sources may be mentioned 1) a differ- 
ence between sexes with females frequently having more and males 
fewer presacral vertebrae; 2) a difference between reciprocal hybrids, 

Vol. 15, No. 3, December 1954 

612 proceedings: symposium on 25 years or 

Table 1. — Distribution of skeletal types in seven inbred strains of mice 


Presacral vertebrae 





A 2 












DBA/2L. . 






23.' 6' 




SEC/2 and 2d 

0. 1 


sometimes matroclinous, sometimes patroclinous ; and 3) a difference in 
frequency of the two asymmetrical types which make up Ai, such that the 
25/26 type occurs two or three times more frequently than 26/25, and of 
the asymmetrical types in A 2 , such that the 26/27 type occurs less frequently 
than 27/26. 

Breeding Experiments 

Nine crosses between selected pairs of strains have been made of the 2 1 
possible crosses between the seven strains (table 2). These 9 crosses 
may be arranged in groups of three, each group involving the same three 
strains. To be specific, 

Group 1 involves 
" 2 " 
u 3 

tt £ u 

" 5 

P, NB, SEC/2 
C3H, DBA, C57BL 
C3H, C57BL, BALB/c 

(Crosses 1, 3, 7) 
(Crosses 2, 5, 8) 
(Crosses 0, 4, 5) 
(Crosses 4, 6, 8) 
(Crosses 0, 2, 6) 

The production of mice is complete in crosses 0, 1, 2, 3, 4, 5, 7, but is 
still in progress in crosses 6 and 8, and in crosses 2A and 3A, which are 
repetitions of crosses 2 and 3. The results of crosses and 2 have been 
described by Green (S) and Green and Russell (4). The first and third 
groups of crosses are therefore the only groups for which all three crosses 
are complete. The data of the first group only are presented in this paper. 

The two strains, denoted by Pi and P 2 , in a given cross were mated 
reciprocally, and the hybrids so produced were in turn mated to produce 
other generations in accordance with the scheme in text-figure 1. All 
of the indicated generations were produced in cross 1, P X SEC/2; Fi, 
F 2 , Bi, and B 2 were produced in cross 3, P X NB; and all but B m and B 222 
were produced in cross 7, NB X SEC/2. Crosses 1 and 7 were designed 
with replications. In cross 1, 10 matings and their reciprocals made up 
by sibs were used to produce the F x generation, and 4 of these blocks 
were systematically expanded through the segregating generations. In 
cross 7, 14 blocks were started, 13 contributed to the Fi generation and 
only one block was successfully expanded through the segregating genera- 
tions. In cross 3, the block system was not used. The experiment 

Journal of the National Caneer Institute 

Table 2. — Schematic diagram of crosses between inbred strains 




Cross number 






SEC/2 and 2d, 



Cross C57BL X BALB/c 
"IP X SEC/2 

" 2 C3H X C57BL 
" 3 P X NB 

" 4 DBA X BALB/c 

Cross 5 DBA X C57BL 

" 6 C3H X BALB/c 

" 7 NB X SEC/2 

" 8 C3H X DBA 

was started with 5 matings of type cfP X 9NB and 5 matings of the type 
cfNB X 9P, with no special attention to matching the reciprocals. 

Pi P 2 


-B r 



P 2 



B 2 - 

B 9 

P 2 


P 2 

Bui B222 

Text-figure 1. — Diagram of mating system for identification of various generations. 

All three of the strains used in crosses in group 1 are short-ear strains. 

The named genes of the three strains are: 

P: aa bb CC dd pp sese 
NB: aa bb c eh c eb dd pp 

QTrrvo. J aa bb c eh c eh dd PP sese 
tiih^/z. | oo bf) ceh( , eh Dd pp sese 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 

The SEC/2 strain contains both dense Dd and dilute dd mice, but only 
dd mice were used in the crosses with P and NB. 


The results of the crosses in group 1 are summarized in tables 3, 4, and 
5. The percentages of mice of the five kinds (25, Ai, 26, A 2 , 27) are 
given for each generation separately. No distinction is shown in the 
tables for the two sexes, for blocks, for reciprocals, for coat colors, or 
for numerous other variables which are necessarily a part of the experi- 
ment. The analysis of the results of these crosses is still in progress. Only 
interpretations based upon the coarse groupings of the data as in tables 3 
to 5 are available at this time. 

Table 3. — Distribution of skeletal types in various generations in cross S, P X NB 





A 2 



P 1 = P 



97. 1 




P 2 =NB 













F 2 





B 2 



Table 4. — Distribution of skeletal types in various 

generations in cross 7, 






A 2 



P 1= NB 



91. 6 







67.' 9' 


13. 6 








P 2 =SEC/2 

6. i 




F 2 


F 3 


B x 





B 2 






For the purpose of describing the results of the crosses, it will be con- 
venient to use the terms low, intermediate, and high to refer to the P, NB, 
and SEC/2 strains, respectively. 

Nongenetic variation. — The three parental generations and the three Fi 
generations all exhibit some phenotypic variability. This variability 
appears to be wholly, or almost wholly nongenetic. The evidence for this 
assertion is that, in the strains and crosses where the data are adequate to 
make the test, the parent-offspring correlations are not significantly 
different from zero. For example, in the high SEC/2 strain, 352 mice were 
produced from matings of the type 27 X 27, and were distributed in the 
classes 26, A 2 , and 27 with frequencies of 45, 61, and 246. Matings of the 

Journal of the National Cancer Inslitute 


Table 5. — Distribution of skeletal types in various generations of cross 1, P X SEC 1 2 



A x 


A 2 



P 1 = P 


0. 1 

1. 1 



20. 1 

95. 1 




P 2 =SEC/2 




Fi . 


F 8 




F 3 







B 2 


















type 26 X 27 produced 27, 36, and 103 mice in the same three classes. 
Altogether, six distinguishable mating- types within the strain produced 
mice in the three classes in frequencies which were approximately propor- 
tional (x 2 = 10.95; x 2 (10; 0.05) = 18.31). Similarly the parent-offspring 
correlations of Fi parents with F 2 or Bi or B 2 offspring were all small and 
nonsignificant in cases which permitted a test. 

Genetic variation. — On the other hand, when parents were from a 
segregating generation and the data permitted a test, the parent-offspring 
correlations were large and significant. This is evidence that genes are 
involved in the differences between the low, intermediate, and high strains. 
The evidence from the backcross generations leads to the same conclusion. 
Particularly in cross 1, the cross of low by high, where first, second, and 
third backcrosses to both parental strains are available, the shift of the 
distributions toward the parental types is easily seen. 


One gene pair model. — The cross of low by intermediate (cross 3, P X NB) 
produced 91.2 percent of mice with 26 presacral vertebrae in the Fi gen- 
eration. The cross of intermediate by high (cross 7, NB X SEC/2) pro- 
duced 91.6 percent with 26 presacral vertebrae. The cross of low by high 
(cross 1, P X SEC/2) produced 97.6 percent with 26 presacral vertebrae. 
These results, taken alone, suggest that the intermediate strain contains 
genes which are dominant both to low and to high strain genes. 

We may test the breeding results against various genetic models if we 
may regard a genotype as fixing a central developmental pattern which 
is realized or not as a constant phenotype, depending upon the amount of 
"blurring" due to nongenetic causes of variation. 

A model of one pair of genes for the difference between the low and inter- 
mediate strains, with a gene for intermediate dominant to its allele for low, 
indicates that the percentages of mice with 26 presacral vertebrae in the Pi, 
P2, Fi, F 2 , Bx, and B 2 generations should be: 0, 100, 100, 75, 50 and 100. 
The observed percentages were 3, 99, 91, 82, 37, and 97. For the difference 

Vol. 15, No. 3, December 1954 

616 proceedings: SYMPOSIUM ON 25 YEARS OF 

between the intermediate and high strains, with a gene for intermediate 
dominant to its allele for high, the model indicates that the percentages of 
26 presacral vertebrae in Pi, P 2 , . . ., B 2 i should be 100, 0, 100, 75, 75, 
100, 100, 50, 25, 75, and 100. The observed percentages were 99, 15, 92, 
80, 79, 97, 97, 55, 19, 70, and 95. 

These same two pairs of genes, each showing dominance of intermediate 
over departures from intermediate may be used to construct a model for 
the cross of low by high. The model indicates that the percentages of 
26 presacral vertebrae in Pi, P 2 , . . ., B 2 i should be : 0, 0, 100, 62, 62, 50, 
25, 12, 50, 25, 12, 75, and 75. The observed percentages were 3,15,98, 
96, 95, 63, 23, 10, 90, 63, 39, 99, and 82. 

;$|The discrepancies between the breeding results and the one gene pair 
model are conspicuous and numerous. If allowance for the nongenetic 
variation is made, the agreement is better but not satisfactory. It seems 
clear that models based on one or two pairs of genes will not be adequate. 

While the agreement is not satisfactory, there is an unmistakable 
tendency for the observed and theoretical proportions to vary together. 
This suggests that the number of pairs of genes is not large and that an 
analysis with a slightly different approach may be profitable. 

Multiple factor genetic model with thresholds. — The formation of the 
sacrum is the consequence of a co-ordinated development of certain dis- 
tinctive structures on three or four vertebrae and on the two innominate 
bones of the pelvic girdle. The first vertebra numbered in sequence from 
the cranium to participate in sacral formation may be the 26th, 27th, or 
28th. It will be convenient to think of the anterior border of the sacrum 
as defined by two points, one right, one left, the locations of which vary from 
mouse to mouse and from side to side of the same mouse. Sometimes the 
points rest on the 26th vertebra, sometimes on the 27th, sometimes on the 
28th, and sometimes on different, but adjacent vertebrae. 

Further, we may assume that if the points lie anywhere on the 26th 
vertebra, whether toward its anterior border or toward its posterior border 
or in the middle, or whether one point is in a slightly different location from 
the other, the 26th vertebra will participate in sacralization, thereby 
leaving 25 presacral vertebrae. Similar assumptions may be made about 
the 27th and 28th vertebrae, leading to 26 and 27 presacral vertebrae, 
respectively. However, if the right and left points lie on different 
vertebrae, the vertebra with the more anterior point becomes, under this 
view, an asymmetrical vertebra since sacralization is exhibited on one side 

The hypothetical points, of either the right or left side, may lie anywhere 
along the 26th, 27th, or 28th vertebra, but the probability for them to lie 
on a given position is a function of the position, such that the probability 
changes from one position to another. If the contrary were the case, 
that is, if the probability were constant from position to position, the 26th, 
27th, and 28th vertebrae would be involved in sacralization equally often. 
This is clearly not the case. Rather the probability for the points to lie on 

Journal of the National Cancer Institute 



a given position is determined by the genotype of the zygote and by the 
environment in which the zygote develops. 

The location of the points for a collection of mice of a given generation 
may be denned by a probability distribution with one or more parameters. 
If the number of genetic and nongenetic factors acting upon the location 
of the points is large, if individually they have relatively small effects, and 
if the mutual intercorrelations of the effects are zero, the probability 
distribution may be expected to be normal in form and specifiable by a 
mean and a variance. 

The probability distributions of the right and left side may both be 
normal in form but have different means and different variances. The 
inequality of the means, of the variances, or of both may account for 
asymmetrical vertebrae. On the other hand, the probability distributions 
of the two sides may have equal means and equal variances, but since the 
correlation between the two sides is not perfect, asymmetrical sacra may 
be merely an expression of imperfect side-to-side correlation. 

The probability distributions for the two sides may be combined into a 
single distribution, if it is assumed that the means and variances are equal. 
Doing so produces five categories of mice: those with 25 presacral verte- 
brae, those with 26, those with 27, those with asymmetrical combinations 
of 25 and 26, and those with asymmetrical combinations of 26 and 27. 
The points which mark the boundaries between the 26th and 27th, and 
between the 27th and 28th vertebrae may be referred to as "thresholds." 

The four thresholds so defined may be denoted as the T, U, V, W 




(=25 presacral 


Ax 27 

(=26 presacral 


(=27 presacral 


All of the scale intervals must of course be estimated from data. 

Test of the multiple jactor model. — The multiple gene model with thres- 
holds may be tested with the data available from the three crosses (tables 
3, 4, 5). The method consists of finding the mean and variance of each 
generation on a scale for which the unit of measurement is a "threshold 
unit." That is, the difference between the thresholds which separate an 
asymmetrical type from the two adjacent symmetrical types may be 
taken as unity. This method was first used by Wright (14). Its use in 
connection with variation in the position of the sacrum was described by 
Green (3). 

For the cross of low by intermediate (cross 3, P X NB), the means and 
variances computed from the data are given in table 6. Under the 
multiple gene model with equal and additive effects of the "plus" alleles, 
with no linkages, and with no nonallelic interactions, there are several 
critical relationships between the means of various generations. For 
example, using m to denote a true mean, x to denote a sample mean, and 

Vol. 15, No. 3, December 1954 



subscripts to denote the generation, the expectations E of the F 2 , Bi, and 
B 2 means are: 

where M=\ (m Pi +m P2 ) ? is the midparent, 

Table 6. — Computed means and variances and expected means in the cross of low by 
intermediate. Cross 8, P X NB. xu=0, xu—x T =l 





P 1= P 






(2. 4)2* 
(2. 4) 2 * 
(2. 4)2* 
(2. 8) 2 
(2. 3) 2 
(2. 3) 2 

P 2 =NB.. 

8. 6t 


F 2 

B 2 

-6. 6 

B 2 

5. 6 

♦Average of Pi and F! variances. 
tEstimated from E{x p )=4w F -2m v -w» p . 

In cross 3, the location of the P 2 mean cannot be computed from the 
observations on the P 2 generation. Its location may be computed from 
E (*p 2 )=4mF 2 — 2m Fi — m Pi , provided the means of Pi, Fi, and F 2 are 
computable from their respective generations. The expected means of 
P 2 , Bi, and B 2 are shown in table 6 for comparison with the observed 
means. The means and variance given in table 6 lead to a construction 
of probability distributions like those in text-figure 2. 

Table 7. — Computed means and variances and expected means in the cross of intermediate 
by high. Cross 7, NB X SEC/2. x v =0, x w -x v =l 


Px = NB... 
P 2 = SEC/2 


F 2 

F 3 



B 2 

B 22 

Bi 2 

B 2 i 






(1. 90)2 
(1. 90) 2 
(1. 90) 2 
(4. Ol) 2 
(4. 01) 2 
(2. 21) 2 * 
(2. 14) 2 1 
(2. 21)2 
(2. 14) 2 1 
(2. 14) 2 1 
(2. 14)2t 

Expected means 

tion 1 



3. 6 

4. 1 

2. 1 

3. 7 

tion 2 



tion 3 

-8. 1 


♦Variance of Bi not computable. Variance of Bj used as approximation, which is satisfactory only if there is 
no dominance. 

fVariances of B33 and B21 averaged, and average used as approximation of variances of Bu and Bu, which is 
satisfactory only if there is no dominance. 

Journal of the National Cancer Institute 



For the cross of intermediate by high (cross 7, NB X SEC/2), more 
generations were available. The means and variances computed from 
the data are given in table 7. To compute the variances, several assump- 
tions were used: 1) that the variances of Pi, P 2 , and F x are equal; 2) that 
the variances of F 2 and F 3 are equal; 3) that the variance of Bi which 
was not computable is equal to the variance of B 2 , an assumption which 
is valid only if there is no dominance; 4) that the variances of B 22 and B 2i , 
being theoretically equal, may be averaged; and 5) that the variances of 
Bn and B i2 , neither of which were computable, could be estimated by the 
average variance of B 22 and B 2 i, an assumption which also is valid only 
if there is no dominance. The location of the mean of Pi is in doubt, 
since over 99 percent of mice in the NB strain had 26 presacral vertebrae. 
To yield this percentage, the mean should be at 4.6 units below the V 
threshold. This leads to a set of expected means given under assumption 
1 in table 7. From the means of P 2 , F x , and F 2 , the P x mean may be esti- 
mated as lying 10.5 units below the V threshold. This estimate is 
unexpectedly large and leads to a set of expected means given under 

t u 


J 1 

MF 2 F, B 2 



B, B 2 

Text-figure 2. — Construction of theoretical distributions for the cross of low by 
intermediate. Cross 3, P X NB. 

Vol. 15, No. 3, December 1954 

316263 — 54 22 



assumption 2 in table 7. A smaller estimate of 8.1 units below V is 
obtained if it is assumed that the Fi, F 2 , and F 3 means are all estimates 
of the same quantity, that is, if it is assumed that there is no dominance. 
This yields the set of expected means given under assumption 3 of table 7. 
The computed means and variances yield a construction of probability 
distributions such as in text-figure 3. 


B n B 21 F 3 
P, P, B, F 2 F, B 12 B 2 



Expected (1) 


p 1 B l F 2 B 2 B22 

Text-figure 3. — Construction of theoretical distributions for the cross of intermediate 
by high. Cross 7, NB X SEC/2. 

For the cross of low by high (cross 1, P X SEC/2), all of the generations 
through third backcrosses to both parental strains were available. The 
means and variances of each generation computed from the data are given 
in table 8. The distance between the V and W thresholds was taken as the 

Journal of the National Cancer Institute 



"threshold unit." The distance from U to V, estimated from the com- 
bined data of the F 2 and F 3 generations, is 11.7 units. The distance be- 
tween the T and U thresholds is 0.7 units. The following assumptions were 
also made: 1) the variances of Pi, P 2 , and Fi are equal; 2) the variances of 
F 2 and F 3 are equal and estimable from the combined data of the F 2 and 
F 3 generations; 3) the variance of B !2 , which was not computable owing to 
the fact that 99 percent of B i2 mice had 26 presacral vertebrae, is the same 
as the B 22 variance, to which it is theoretically equal. These computations 
and assumptions yield the construction of probability distributions for the 
various generations of cross 1 shown in text-figure 4. 

The minimal number of effective factors which distinguish the strains 
may be estimated from the ratio of the square of the difference between the 


1 B m B 11 B l B 21 F i F 2 B : 

L i 1 ti LJLJ 


*m B n B i 


B 22 B 222 P 2 


Text -figure 4. — Construction of theoretical distributions for the cross of low by hiqh 
Cross 1, P X SEC/2. 

Vol. 15, No. 3, December 1954 



Table 8. — Computed means and variances and expected means in the cross of low by 
high. Cross 1, P X SEC/2. x v =0, x w -x v =l, xy-x v =11.7 





Pi = P 

P 2 = SEC/2 


F 2 

F 3 




B 2 


















(1. 78) 2 
(1. 78) 2 
(1. 78) 2 
(2. 88) 2 
(2. 88) 2 
(1. 54) 2 
(1. 17) 2 
(1. 20) 2 
(4. 15) 2 
(2. 69) 2 
(1. 85) 2 
(2. 69) 2 * 
(1. 91) 2 



0. 1 




11. 1 




*Variance of B i2 not computable, 
used as an approximation. 

Variance of B 2 2 to which it is theoretically equal 

parental means and the additive genetic variance [Castle (15); Mather 
(16) ; Wright (17)]. The P and NB strains appear to differ by at least 4 
pairs of genes, NB and SEC/2 by at least 1 pair, and P and SEC/2 by at 
least 21 pairs. 

Dominance is indicated by the failure of the F 2 mean to lie on the same 
point as the F x mean. In cross 3 (low by intermediate), the F 2 mean is 
slightly below the F x mean, indicating possible dominance of intermediate 
over low. In cross 7, there is evidence of dominance of high over inter- 
mediate. In cross 1, the relationship is reversed, low being partially 
dominant to high. 


A multiple gene pair model is very flexible because of the large number 
of parameters to be estimated from the data. These include the relative 
contributions of genetic and nongenetic factors to the total variability, the 
number of pairs of genes or of effective blocks of genes, the average rela- 
tionship between alleles (dominance) , and the average relationship between 
nonalleles (epistasis) . 

For variations in the position of the sacrum, the multiple gene pair 
model has been used to provide a concept of a continuous underlying vari- 
able which is manifested only in 2, 3, 4 or 5 categories separated by thres- 
holds. The scale for the underlying variable has been chosen deliberately 
to minimize the nonallelic interactions, and to provide a scale for displaying 
additive effects of nongenetic and genetic causes of variation, including 
dominance. The agreement between the observed results and the genetic 
model is clearly not perfect, and at a few points is not even good. Never- 
theless a genetic model based upon the concepts of quantitative inheritance 
appears to be more promising than more elementary models for explaining 
variation in the position of the sacrum. 

The development of a given mouse is such that the sacrum forms from 
vertebrae following the 25th, 26th, or 27th vertebra. Many agents govern 

Journal of the National Cancer Institute 



the particular position of the sacrum. These agents cannot be specified 
beyond saying that some are genetic and some are nongenetic. The inter- 
play of the genetic and nongenetic agents appears to be such that the 
position of the sacrum for a given mouse cannot be predicted with cer- 
tainty. Rather, by use of a multiple gene model, it is possible to specify 
the probabilities for each of several sacral positions for mice of a given 
strain or for mice of a given generation in crosses between strains. 

The model also suggests that the effects of experimental modifications 
of the embryonic environment will not be exhibited in qualitative changes 
in the individual embryos. Rather the detectable effect, short of the 
production of abnormalities, will be a change in the frequency distribution 
of skeletal types. This means that the set of probabilities of the various 
types, rather than the types themselves, are changed with the changing 
environment. This has been borne out in experiments with X irradiation 
of embryos [Russell and Russell (18)] and in experiments using ova 
transplantation [Green and Green (19)]. 


(1) Green, E. L.: Genetic and non-genetic factors which influence the type of the 

skeleton in an inbred strain of mice. Genetics 26: 192-222, 1941. 

(2) Searle, A. G.: Genetical studies on the skeleton of the mouse. IX. Causes 

of skeletal variation within pure lines. J. Genetics 52: 68-102, 1954. 

(3) Green, E. L.: The genetics of a difference in skeletal type between two inbred 

strains of mice (BalbC and C57blk). Genetics 36: 391-409, 1951. 

(4) Green, E. L., and Russell, W. L.: A difference in skeletal type between recip- 

rocal hybrids of two inbred strains of mice (C57BLK and C3H). Genetics 36: 
641-651, 1951. 

(5) Green, E. L., and Green, M. C.: The effect of uterine environment on the 

skeleton of the mouse. Jour. Morph. 78: 105-112, 1946. 

(6) Dobrovolskaia-Zavadskaia, N.: Sur la mortification spontanea de la queue 

chez la souris nouveau-n6e et sur Texistence d'un caractere (facteur) hereditaire 
"non-viable." Compt. rend. Soc. biol. 97: 114-116, 1927. 

(7) Chesley, P., and Dunn, L. C.: The inheritance of taillessness (anury) in the 

house mouse. Genetics 21: 525-536, 1936. 

(8) Gluecksohn-Schoenheimer, S.: The morphological manifestations of a domi- 

nant mutation in mice affecting tail and urogenital system. Genetics 28: 
341-348, 1943. 

(9) Dunn, L. C., and Gluecksohn-Schoenheimer, S.: Stub, a new mutation in the 

mouse with marked effects on the spinal column. J. Hered. 33: 235-239, 1942. 

(10) MacDowell, E. C., Potter, J. S., Laanes, T., and Ward, E. N.: The manifold 

effects of the screw tail mouse mutation. J. Hered. 33: 439-449, 1942. 

(11) Green, E. L., and Green, M. C.: Effect of the short ear gene on number of ribs 

and presacral vertebrae in the house mouse. Amer. Nat. 80: 619-625, 1946. 

(12) Carter, T. C.: The genetics of luxate mice. I. Morphological abnormalities of 

heterozygotes and homozygotes. J. Genetics 50: 277-299, 1951. 
(IS) Green, M. C.: A new inherited leg and foot abnormality, luxoid, in the house 
mouse. (Abstract.) Genetics 38: 666, 1953. 

(14) Wright, S.: The results of crosses between inbred strains of guinea pigs differing 

in number of digits. Genetics 19: 537-551, 1934. 

(15) Castle, W. E.: An improved method of estimating the number of genetic factors 

concerned in cases of blending inheritance. Science 54: 223, 1921. 

Vol. 15, No. 3, December 1954 

624 proceedings: symposium 

(16) Mather, K.: Biometrical genetics. New York, Dover Pub., Inc., 1949, p. 158. 

(17) Wright, S.: The genetics of quantitative variability. In Quantitative Inheri- 

tance. London, Her Majesty's Stationery Office, 1952, pp. 5-42. 

(18) Russell, L. B., and Russell, W. L.: Changes in the relative proportions of 

different axial skeletal types within inbred strains of mice brought about by 
X-irradiation at critical stages in embryonic development. (Abstract.) Ge- 
netics 35: 689, 1950. 

(19) Green, E. L., and Green, M. C.: Modification of difference in skeletal types 

between reciprocal hybrids by transplantation of ova in mice. (Abstract.) 
Genetics 38: 666, 1953. 

Dr. P. B. Sawin, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

The author is to be commended for the masterful demonstration of one of the 
basic attributes of homeotic variations, the origin of which has for more than three 
quarters of a century been a puzzle to many prominent biologists, including Bateson, 
Kingsley, Kingsbury, Bumpus, Todd and Keith et al. These leading morphologists 
of their time attempted to solve the problem from the phylogenetic point of view, but 
the two theories proposed have never been substantiated experimentally. 

Dr. Green's insight as to the best sort of genetic material and his thoroughness in 
carrying out established genetic procedures have led to the most clear-cut conclusions 
yet reached with regard to this sort of variation. It was Danforth, I believe, who first 
suggested a genetic interpretation, and there have been a number of attempts to inter- 
pret the inheritance of such variation in terms of one or more gene substitutions. 
None, however, have been found which survive thorough genetic investigation. This 
is particularly significant, I think, in view of the known single-gene effects upon such 
variations in species such as Drosophila and Habrobracon and upon the extremities 
(limbs and tail) in mammals. Why single genes should be adequate for such variations 
in the lower forms and not in mammals, and why in the mouse single genes adequately 
explain the inheritance of the numerous postsacral but rarely presacral variations 
are moot questions. 

With respect to the first question, it may also be of significance that although in 
Drosophila such variations tend to be mono-segmental, there is evidence from the 
work of Villee, at least, that in the case of certain genes more than one adjacent seg- 
mental part is frequently affected (pleiotropically) . Perhaps the fact that the gene 
effect is not more generalized as it is in mammals could be due to the relatively lower 
growth potential of the species or to the relatively small number of segments (low 
segmental rate). With regard to the second, the question may well be asked: "Is it 
because these genes are terminal in their effect that their inheritance is more precise, 
whereas the inheritance of homeotic variations, which are sub terminal in development, 
may become complicated by the very nature of relations with their neighboring parts?" 
The fact that Dr. Green has found the position of the pelvis and sacrum can be governed 
by many agents both genetic and environmental, plus the fact that it can be further 
complicated by the interplay of these agents, tremendously increases the complexity 
of the problem. Yet Dr. Green seems to have developed a system from which pre- 
dictable results can be obtained. 

In the rabbit we have gone about the same problem in a somewhat different manner. 
Stimulated by the genetic studies of Fisher and Kiihne in man and the rabbit, in 
which they have stressed the importance of relationships which may exist between 
more than one homeotic variation, we have focused our attention upon two races in 
which there is a high correlation between the variations at two borders, the thoraco- 
lumbar and lumbosacral, the one race having both extra units in 95 percent of cases, 
the other lacking them. Quantitatively the increase in relative growth between these 
two borders is demonstrable in the adult, as well as at birth, and more recently we have 
found it in 21-day embryos. 

In the left half of figure 1, a representative specimen of race III, we see the enlarged 
lumbar growth area characteristic of that race. It is manifested in the greater size 
and number of lumbar centra at 21 days. In the right half of figure 1, showing a repre- 
sentative of race X, we see the enlarged anterior growth area manifested by the greater 
size and number of the elements of the neural arch at the same age. These two speci- 
mens, however, are far from portraying the entire situation. As Dr. Green has so well 
shown, one cannot expect to find the effects of experimental modifications of the 
embryonic environment exhibited in individual embroyos, and this is also apparent 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

626 proceedings: symposium 

with respect to the growth forces acting on these embryos at this time. I have chosen 
here two specimens representative of the average pattern. Time will not permit of 
my demonstrating details of these points at this time. In fact, the evidence is still not 
complete. What we have found can be summarized in this way. Considering these 
two races as a whole there are apparent four characteristic differences between them 
which somehow interact to produce threshold differences at various levels of the 
axial skeleton. At this age they are 1) a tendency to generalized precosity of onset of 
ossification in race X plus certain localized differences anteriorly; 2) genetic influences 
which do not act on all localized organs simultaneously; for example, neural arches are 
quite independent of the centra; 3) growth in width of centra is more precocious than 
growth in length, and especially so in X. There is evidence that it is changing very 
rapidly and is not relatively comparable to adult width at this time; 4) the regional 
growth areas as a whole are in a much greater state of flux at this time than at birth or 
in the adult. 

Differential growth of parts in relation to each other determines essentially the 
characteristics which define the kind of vertebra. We suspect that the clue to the 
specific gene actions lies in such differences which in turn interact to determine the 
thresholds of homeotic change described so clearly by Dr. Green. Whether these 
differences can be traced back through cartilage formation, through mesenchyme con- 
densation and through still earlier developmental processes is a question which can 
only be answered by embryological techniques and measured by statistical methods. 
We have seen them at three developmental stages. Between these stages it would seem 
to be only a matter of time and effort to describe the gene influence involved. Here the 
more refined, longitudinal approach now being used by the physical anthropologists is 
particularly important. The well known adolescent spurt in man is one example 
where acceleration or retardation of the growth of the whole or the part in relation to 
the whole is capable of profound changes in the end result. It is my belief that the 
final analysis of the problem of the genetic and environmental basis of homeotic 
variations must ultimately come from the combined efforts of the embryologist and the 
geneticist who have command of adequate statistical techniques and who approach 
the underlying growth processes or gradients by longitudinal methods. Since longi- 
tudinal methods are not yet possible embryologically, the importance of inbred strains 
such as those of this laboratory cannot be overemphasized. 

Plate 44 

Figure 1. — Comparison of anterior and lumbar growth areas of 21-day embryos of 
Race III (left) and Race X (right) from cleared specimens shown at the same 







i ' 

. 4*1 


iv "I 


Figure 1 


316263—54 23 

Genetic Control of Embryonic Growth 
and Differentiation 1,a 

Salome Gluecksohn-Waelsch, Department of 
Obstetrics and Gynecology, College of Physicians 
and Surgeons, Columbia University, New York, 
N. Y. 

The identification and the analysis of the different factors which co- 
operate intimately with each other to produce a normal organism is one 
of the most important goals of all students of development. That these 
factors are of varying origin, and are in turn subject to various deter- 
mining influences, is of course well known and does not need to be stressed 
here. The recognition and assignment of developmental factors to 
various categories, e.g., genetic, environmental, etc., are prerequisites 
for a study of the mechanisms by which they exert their control of de- 
velopment. In this paper, I should like to review present knowledge of 
some genetic factors which have been found to control processes of em- 
bryonic growth and differentiation. The organism in which these par- 
ticular factors were studied is the house mouse, which plays such an 
important role in the past and present history of this laboratory. But 
before I begin the discussion of this material, I should like to give an 
illustration taken from a considerably lower creature than the mouse, of 
how various factors may interact in the operation of developmental 
mechanisms. I have always been particularly intrigued by the outcome 
and the interpretation of one of Spemann's most interesting experiments. 
Conceived as early as 1921 by Spemann the experiment was actually 
carried out by Schotte (1) in 1932. The question was asked if the re- 
acting system of a certain species would respond to an inductive stimulus 
of a different species in its own species specific way or in the pattern 
of the inducing species. The experiment consisted in the transplantation 
of belly ectoderm from an anuran species to the mouth region of a urodele 
embryo — a so-called xenoplastic transplant. This ectoderm that nor- 
mally would have formed skin of the abdomen of a frog, formed structures 
of the mouth region in response to the inductive stimulus provided by the 
mouth environment of the salamander host. The environment thus 
directed the differentiation of the transplanted tissue into channels 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine 
June 28, 1954. 

» This work was supported in part by a research grant (G-3676) from the National Institutes of Health, U. S 
Public Health Service, and in part by a grant-in-aid from the American Cancer Society upon recommendatio 
of the Committee on Growth of the National Research Council. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

630 proceedings: SYMPOSIUM ON 25 YEARS of 

conforming with the environment's demands. However, the hereditary 
make-up of the transplanted tissue manifested itself also — namely, in 
the type of mouth structures formed by the implant: these were suckers, 
typical of tadpoles of anuran species in contrast to the balancers normally 
formed by the host. The implant thus retained its genetically controlled 
species characteristics while giving in to the demands of the environment 
of the host in respect to regional differentiation. 

The interaction of heredity and environment in the formation of a 
developmental pattern seems to me beautifully illustrated by this example. 

For the detailed analysis of the control of developmental patterns by 
particular genetic factors amphibians do not offer the most suitable 
material mainly for reasons connected with the length of their life cycle. 
In this respect the mouse is much more cooperative, as you all know, 
with the speed and quantity of its reproductive activity. It is because 
of this reason, as well as because of the existence of a number of mutations 
concerned with processes of early growth and differentiation, that today 
we know perhaps more about the specific role of genes in development of 
the mouse than of any other vertebrate. 

As mentioned before, developmental patterns owe their origin to various 
genetic as well as nongenetic factors. Among the genetic factors there 
are not only those more or less directly concerned with the operation of 
particular processes of differentiation, but a] so those which determine 
the type of reaction to genetic and nongenetic influences to which the 
organism is subjected during its development. 

The example which I am about to mention will serve to illustrate 
typical abnormalities of a developmental pattern which may be brought 
about either by changes of genetic or of nongenetic factors; the mani- 
festation of these factors in turn depends to a large degree on the indi- 
vidual's genetically controlled susceptibility. 

The abnormality I want to talk about is cleft palate where a cleft 
extends through the premaxilla and the entire length of the hard palate: 
this disturbance of a normal developmental pattern is found rather 
frequently in mammals. The cleft seems to be the consequence of the 
failure to unite of the two palatine processes of the maxillae and traces 
back perhaps to a reduced growth rate of the maxillary processes. 

It is interesting that hereditary harelip or cleft palate has been re- 
ported several times in the literature but never with a simple genetic 
basis (#). Some years ago we described the occurrence of one type of 
cleft palate which owed its existence not to the effect of one or more 
specific genetic factors but rather to the combination of a special geno- 
type and a peculiar set of environmental conditions (3). In this case 
intimate interaction of a set of genetic and environmental factors was 
responsible for the disturbance of the normal developmental pattern of the 
palate. On the other hand, cleft palate as a regular symptom in a whole 
syndrome of abnormalities has been observed in our laboratory twice in 
recent years; two different recessive factors in homozygous condition 
were responsible for it. Each of these recessive factors produced its own 

Journal of the National Cancer Institute 



syndrome of abnormalities: in the case of one (pc) various effects were 
observed in other parts of the skeleton as well, particularly in the extremi- 
ties, while in the case of the other one (ur) abnormal functioning of the 
kidneys resulting in water retention, in addition to other skeletal abnor- 
malities, occurred in the homozygotes. Thus cleft palate seems to mani- 
fest itself more regularly in combination with other abnormalities. 

Cleft palate may also be produced experimentally in mice by the use 
of different deleterious agents. Fraser {4), for example, reported as 
much as 80 percent of cleft palate in genetically normal offspring of 
certain strains whose mothers had been given cortisone. Here, also, the 
incidence of the abnormality depended largely on the genetic makeup 
of the treated strain. 

It thus seems that growth and differentiation of the embryonic material 
forming the hard palate are subject to many influences, genetic and 
nongenetic. There are a number of reasons which make it likely that 
these events take place rather late in development. 

Now I should like to discuss some material which deals with develop- 
mental mechanisms operating considerably earlier in the life of the 
embryo. One of the organ systems of the developing embryo most sus- 
ceptible to all types of changes, and reacting to them with abnormalities 
of different kinds and degrees, is the nervous system. A great many 
genetic factors, as well as several environmental agents, are known to 
cause disturbances in the normal development of this important organ 
system in the mouse. As you all know, the nervous system shows a 
particularly close relationship to other developmental systems of the 
embryo during its morphogenesis; its complete dependence on the under- 
lying notochord-mesoderm material for its initial differentiation is a well 
known fact, at least in amphibian embryos. Minor variations in 
different embryonic regions during development therefore easily express 
themselves in abnormalities of the highly sensitive nervous system. 

In the mouse, growth and differentiation of the nervous system seem 
to be influenced by a number of genetic factors operating through differ- 
ent channels. One type of brain abnormality resulting from develop- 
mental disturbances is known as pseudencephaly; in this condition the 
bony cranial roof is absent and parts of the brain are inverted, forming a 
cap on top of the head ; failure of the brain folds to close along the dorsal 
midline is connected with this abnormality. Genetically, pseudencephaly 
has been described as the result of a recessive mutation by Bonnevie (5) . 
It also arises in the presence of X-ray-induced chromosomal translocations 
as shown by Snell and coworkers (6) . Abnormalities similar to pseuden- 
cephaly have been investigated recently by Auerbach (in press) in mouse 
embryos homozygous for the dominant mutation Splotch. In our own 
laboratory we have observed a fairly high incidence of pseudencephaly 
among embryos of so-called normal mouse strains. 

The normal inductive relationship between notochord-mesoderm ma- 
terial and the developing nervous system has been shown to be affected 
in the presence of the mutations at the T-locus of Chromosome IX of 

VoL 15, No. 3, December 1954 

632 proceedings: symposium on 25 years of 

the mouse. Abnormalities of the neural tube in the presence of the 
mutation T were traced back by Chesley (7) to those of the notochord. 
Duplications of neural folds in embryos homozygous for Kink, another 
dominant mutation in Chromosome IX, seem to indicate abnormal 
functioning of the material responsible for formation of a single normal 
embryonic axis. In our laboratory recently, Dr. Karl Theiler studied 
the developmental effects of embryos homozygous for Fused, another of 
the dominant mutations of this group, and found multiple neural tubes in 
different regions of the developing embryos. While in this latter case 
the causal role of the notochord-mesoderm material is not clearly indicated, 
studies of other mutations in Chromosome IX point strongly to a general 
effect of the factors in this chromosome on this important developmental 
system. Since the effects of these mutations have been described and 
discussed repeatedly in recent years (8, 9) I do not plan to go into more 
detail about them here. I only want to state in summary, that genetic 
control of embryonic growth and differentiation is most strongly demon- 
strated by the existence of the mutations in Chromosome IX of the mouse 
and their effects on notochord-mesoderm material and the developing 
nervous system. 

The recent investigations by Auerbach of the homozygous effect of 
Splotch, mentioned above, have shown the interference of this mutation 
with the normal functioning of the neural crest material, a significant 
part of the nervous system in embryonic differentiation. 

Another example of a disturbance of normal inductive relationships in 
the developing nervous system by a mutant genetic factor may be found 
in the effect of Kreisler, a mutation in the mouse studied by Paula Hertwig 
(10) . Here, separation of the ear vesicle primordium from the neural tube 
interferes with the normal differentiation of the ear vesicle. 

Numerous further examples of the interference of mutations with normal 
processes of growth and differentiation in other organ systems of the 
mouse could be cited. The question may, of course, be asked whether 
the control of normal growth and differentiation by genes may be inferred 
from the demonstration of abnormal processes of differentiation in the 
presence of mutations. I would tend to answer this question in the 
affirmative and say that one may conclude at least that normal genetic 
factors participate in the control of these same processes shown to be 
abnormal in the presence of mutations. Of course the mechanisms by 
which normal or changed genes exert their control on growth and differ- 
entiation offer separate problems. It is the investigation of these mechan- 
isms which occupies us most at the present and I should like to mention 
briefly a few of the possible approaches to these problems. 

One of them is based on a concept whose origin as well as elaboration 
we owe to Richard Goldschmidt (11). It is the concept of the phenocopy, 
which has been used by Goldschmidt and others to attack problems of 
gene action. If it were possible with the help of an environmental agent 
to reproduce the individual steps in a chain of processes begun by a 
mutant gene, a true phenocopy would result which would be of great 

Journal of the National Cancer Institute 



value for an analysis of the detailed mechanism by which a gene pro- 
duces its phenotypic effect. 

In this connection the abnormalities in the rat reported by Gillman 
and coworkers (12) in the offspring of mothers treated with trypan blue 
seemed to resemble, at least in their final manifestations, the abnormalities 
caused by genie action of mutants of Chromosome IX of the mouse. In 
order to study this resemblance more closely, Hamburgh (13) carried out 
an investigation directed towards an analysis of the individual steps by 
which the abnormalities induced by trypan blue arose. His results showed 
that these were not mediated by the notochord-mesoderm system and 
were thus different from the mutant effects, not being true phenocopies. 

Another approach to the problem of gene action in differentiation 
which we are following up at the moment is based on ideas which conceive 
of the processes of embryonic growth and differentiation as essentially 
similar to those operating in antibody formation. Furthermore, since in 
the case of the mutations in the mouse described above indications exist 
that primary gene action may involve immune reactions, it occurred to us 
that the interference of these mutations with normal processes of growth 
and differentiation might be mediated by a mechanism resembling 
immune reactions. 

In order to investigate this possibility we started out with an exami- 
nation of the effect which immunization of normal mothers against specific 
organ tissues might have on the development of the corresponding organ 
system of their embryos. Indications of specificity of effect were obtained 
from such experiments; immunization with brain affected the developing 
nervous system of embryos more specifically than did immunization with 
heart, which resulted in different types of abnormalities, including those of 
the mesoderm. 

These experiments are in a preliminary state and require extension and 
corroboration. Eventually the study will be extended from normal strains 
to tissues from mutant individuals and both the immunizing effect of 
mutant tissues and the reaction of mutant embryos to immunization of 
the mothers will be investigated. 

From the material available I have selected here for discussion some of 
the studies which demonstrate the role of genes in processes of embryonic 
growth and differentiation in a mammal. On the other hand, we all know 
that these processes are subject to the effect of other than genetic factors 
as well and that genetic and nongenetic factors interact intimately in 
mechanisms of growth and differentiation. The study of genetic control 
of early embryonic processes has opened up a number of interesting prob- 
lems in addition to the descriptive analysis of gene effects. One of these 
is the relative scarcity of genetically caused abnormalities known to occur 
in early development; this might possibly be the result of the considerable 
regulatory power which the mammalian embryo has been shown to possess 
like other vertebrates and which enables it to survive unharmed adverse 
genetic or environmental effects. It is interesting to consider the possible 
genetic control of this regulatory ability which diminishes as development 
proceeds; the similarity of genetic factors controlling regulatory power 

Vol. 15, No. 3, December 1954 

634 proceedings: symposium 

to modifying genes, and the possible mechanism by which they exert their 
control, have been discussed by us recently (9). 

Another interesting aspect of genetic factors controlling growth and 
differentiation is their pleiotropic effect. Frequently, the association of 
different effects as the result of abnormal genie action is not easily under- 
stood on the basis of existing knowledge of the interrelationships of de- 
velopmental structures; such an association might point toward 
hitherto unknown interdependencies of developmental systems which 
thus may become revealed by pleiotropic gene effects. However, it is 
necessary here to remember the limitations imposed by the use of the 
purely descriptive approach to such problems; disregard of these limita- 
tions has frequently produced unwarranted conclusions, for example, in 
regard to gene action from studies of pleiotropic gene effects. 

The study of developmental effects of mutations in the mouse has 
served to confirm in mammals the existence of developmental mechanisms 
demonstrated previously by experimental means in other vertebrates. 
This fact has been stressed repeatedly and will not be discussed here again. 

Of course, the main question encountered in all approaches to problems 
of genetic control of differentiation is that of the mechanism of differentia- 
tion, knowledge of which would no doubt help to clarify problems of gene 
action; on the other hand, it may not be too presumptious to expect that 
the study of genetic factors controlling differentiation will, in its turn, 
contribute to the clarification of the very problem of differentiation. 


(1) Spemann, H., and Schotte, O.: tJber xenoplastische Transplantation als Mittel 

zur Analyse der embryonalen Induktion. Naturwiss. 20: 463-467, 1932. 

(2) Steiniger, F. : Neuse Beobachtungen an der erblichen Hasenscharte der Maus. 

Ztschr. menschl. Vererb.-u. Konstitutionslehre 23: 427-462, 1939. 

(3) Gluecksohn-Schoenheimer, S., and Dunn, L. C: A new type of hereditary 

harelip in the house mouse. Anat. Rec. 102: 279-287, 1948. 

(4) Fraser, F. C, and Fainstat, T. D.: Production of congenital defects in the 

offspring of pregnant mice treated with cortisone. Pediatrics 8: 527-533, 1951. 

(5) Bonnevie, K.: Pseudencephalie als spontane recessive (?) Mutation bei der 

Hausmaus. Norske Vidsk.-Akad. Oslo Skr. Kl. 1, #9, 1936. 

(6) Snell, G. D., Bodemann, E., and Hollander, W.: A translocation in the 

house mouse and its effect on development. J. Exper. Zool. 67: 93-104, 1934. 

(7) Chesley, P.: Development of the short-tailed mutant in the house mouse. J. 

Exper. Zool. 70: 429-459, 1935. 

(8) Gluecksohn-Waelsch, S.: Physiological genetics of the mouse. Advances 

Genet. 4: 1-51, 1951. 

(9) : Lethal factors in development. Quart. Rev. Biol. 28: 115-135, 1953. 

(10) Hertwig, P.: Die Genese der Hirn- und Gehororganmissbildungen bei rontgen- 

mutierten Kreisler-Mausen. Ztschr. menschl. Vererb.-u. Konstitutionslehre 
28: 327-354, 1944. 

(11) Goldschmidt, R. : Physiological Genetics. New York, McGraw-Hill Book Co., 

Inc., 1938. 

(12) Gillman, J., Gilbert, C., and Gillman, T.: A preliminary report on hydro- 

cephalus, spina bifida and other congenital anomalies in the rat produced by 
trypan blue. South African J. M. Sc. 13: 47-90, 1948. 
(IS) Hamburgh, M.: Malformations in mouse embryos induced by trypan blue. 
Nature 169: 27, 1952. 

Dr. Morris Smithberg, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

Dr. Waelsch has brought out many interesting problems in her excellent review 
concerning genetics in development, any one of which could be discussed at con- 
siderable length. However, underlying these problems I believe there are two 
fundamental questions. The first question would ask whether embryonic growth is 
at all under the control of genetic factors. Most biologists, if confronted with this 
question some years back, would have said, "Yes, we would probably expect it to be." 
The evidence for this expectation still had to be uncovered, and Dr. Waelsch has 
presented the affirmative data. However, it it necessary to add that these genetic 
factors can express themselves within the limits of the environment and, likewise, 
that environmental factors can express themselves within the limits of the genetic 
constitution of the organism. 

The second question is, how does the gene act in the control of developmental 
processes? This is a doubly perplexing situation since the underlying mechanisms of 
both gene action and development are as yet not completely understood. 

There are many approaches to these perplexing problems. I will touch on just one 
or two. The use of mutant strains of mice in an attempt to understand gene action 
and development has of necessity been limited to study of processes rather late in 
development. Mutants of pre-implantation stages are very scarce. Those mutants 
which find expression late in development are mostly of a grossly morphologic nature 
and eventually are interpreted through knowledge gained by experimental embryolo- 
gists using lower vertebrates, such as the frog or chick. For instance, the action of a 
mutant gene of the tailless series can be simulated by extirpation of the terminal 
portion of the notochordal anlage of the frog. Perhaps the most notable exception to 
the morphologic-type mutant deals with the work of Dr. Russell of the Jackson 
Laboratory. Dr. Russell has traced expression of a mutant gene to a faulty synthesis 
in the hemoglobin molecule. It is with work on mutants such as this that the investi- 
gator probably gets closest to the actual nature of the gene action. 

The immunological approach to the primary nature of the gene action, as mentioned 
by Dr. Waelsch, is probably the most promising and perhaps needs more elaboration. 
The impetus for research of antigen-antibody-like reactions in development results 
from two papers published simultaneously in 1947 by Tyler and Weiss. Since these 
stimulating papers many investigators, notably Cooper, Spar, Clayton and Flickinger, 
using amphibians, and Schechtman and his students, using the chick, have found 
ample evidence for the epigenesis of antigens during development. In addition, 
Ebert, using chick embryos, has been able to block the differentiation of chick-brain 
and chick-heart selectively by the use of anti-heart and anti-brain sera. Ebert has 
also been able to modify the growth of embryonic chick spleen by placing adult 
chicken-spleen tissue upon the chorio-allantoic membrane. 

It was very interesting indeed to learn that Dr. Waelsch has been successful in 
producing selectively anomalous embryos in inbred strains of mice. The genetic 
uniformity of her material may perhaps give her a more defined type of anomaly. 
Unfortunately, she is not here to give us the results in more detail. 

Along these same lines it was our aim to try to modify the-growth of the regenerating 
liver of an adult mouse with the use of anti-liver serum prepared in rabbits. The 
ability of the liver to regenerate after 2 of the 3 lobes are surgically removed is one of 
the few regenerative processes still maintained in the adult. By the 7th postoperative 
day the lobe left behind usually increased in size closely approaching the total original 
weight at the time of surgery. The increase in weight involved both cell number and 
ceil size. Anti-liver serum or normal rabbit serum, injected after surgery, did not 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

636 proceedings: symposium 

significantly influence the final weight of the growing liver as we had expected. How- 
ever, the anti-serum did modify the mitotic pattern of the regenerating liver. In an un- 
treated control animal, or one in which normal serum was injected into partially 
hepatectomized animals, one finds a profound increase of mitoses at 2 to 3 days and a 
considerable dropping off at day 4 to day 7 after surgery. On day 7 there are practi- 
cally no mitoses to be found. Partially hepatectomized animals injected with anti- 
liver serum show the same increase of mitoses on day 2 to 3 as the controls. However, 
at day 7 the animals receiving anti-liver serum maintained a high level of mitotic 
figures. Since the experiment is only in a preliminary stage, we cannot say much more 
about this discrepancy in mitotic pattern. 

I should like to end this discussion with some general impressions. There was a 
time when the geneticist camped in the nucleus, and the embryologists camped in the 
cytoplasm. Both were very shy, but nevertheless mutually attracted. Only at 
metaphase, with the breakdown of the nuclear membrane did they temporarily meet. 
The embryologists emerged from the other side of the barrier during xenoplastic 
transplants. But as nature would have it, genetics and embryology have finally 
merged, and a better knowledge of development should be the offspring of this union. 
It is the hope of all geneticists and embryologists that this offspring will exhibit 
extreme hybrid vigor. 

Inheritance of Susceptibility to Con- 
genital Deformity — Embryonic Insta- 
bility 1 ' 2 

Meredith N. Runner, Roscoe B. Jackson 
Memorial Laboratory, Bar Harbor, Maine 

The proportion of medical practice and hospital facility devoted to the 
developing baby (e.g. our children's hospitals) is commensurate with that 
devoted to some of our more serious diseases. Infant mortality during 
the past 40 years has decreased from 12 to 2 percent. Of 6 infants that in 
1910 would have died, today 5 of them attain their first birthday. These 
medical achievements, however, are not without their complications. 
Although nutritional deficiencies and infectious diseases are almost non- 
existent in mothers during pregnancy, the prediabetic condition, for 
example, has become increasingly associated with production of abnormal 
babies. Why deformity is associated with diabetic mothers is not under- 
stood, but in general one can say that improved maternal, prenatal health 
has increased the probability of survival of deformed embryos that 
formerly would have succumbed under less favorable circumstances. Ac- 
complishments with prenatal health of the mother have progressed to the 
point where researchers can now consider factors that influence the health 
of the embryo. 

This report will attempt to demonstrate the extent to which we can 
experimentally unravel interactions of genetic and environmental factors 
that influence health of the embryo. My primary objective will be to 
arrive at a plan that will approach an explanation for or even manipulation 
of differences between normal and abnormal morphogenesis. Any plan 
for experimentation must be based upon a combination of pre-existing 
information mixed with a few good assumptions. Since no hypothesis 
is any better than its worst assumption, I will outline the mixture of fact 
and assumption that we may call principles basic to this report. 

Each individual is the resultant of interaction between effects of his hereditary 
background and contingencies of his environment. In other words each of us 
is the product of nature, in a genetic sense, and nurture, in the broad 
environmental sense. Your presence here today is ample proof that nature 
and nurture have not had a major clash since the time of your own con- 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 28, 1954. 

a These investigations were supported in part by a research grant G-3859 from the National Institutes of Health, 
Public Health Service, U. S. Department of Health, Education, and Welfare. 


Journal of the National Cancer Institute, 
316263—54 24 

Vol. 15. No. 3, December 1954 


Few morphologic characteristics provide geneticists with an a priori basis 
with which to foretell the respective importance of heredity and environment. 
We know that when all the evidence is tabulated and lenient extrapola- 
tions are made, considerable proportions of individual differences in 
morphology are attributable to environmental, nongenetic factors. Since 
no genie machinery operates in a vacuum, it becomes essential to investi- 
gate the environment with which heredity reacts. If we cannot have good 
heredity let us console ourselves by the fact that susceptibility to some 
hereditary defects can be offset by superior environment. 

Infinitesimally small visible variations are important during early phases 
of prenatal development. Although these early minute variations are often 
differences of degree and difficult to quantitate, the embryo has limited 
corrective powers and these variations may account for major differences 
between you and the hopelessly deformed individual. 

Any given process of normal development is regulated by numerous genie 
units and the primary action of any one gene is probably many steps removed 
from direct control of normal embryology. Embryologic determination may 
be the result of several genes acting concomitantly and others acting 
sequentially. Grimeberg (1) has suggested an analogy with evolution 
which is supposed to be irreversible though it is brought about by genie 
mutations each of which is reversible. High stability of tissue interactions 
during normal development may offer a selective advantage. Determina- 
tion probably happens in stages which successively narrow the develop- 
mental possibilities of a region until it finally becomes completely deter- 
mined. The ultimate morphologic results are determined by a series of 
concomitant and preceding gene effects on a biochemical level. 

An evolutionary safeguard for preservation of the species is to have the 
fate of the embryo independent of any one single, critical, genetic locus. 
On the other hand, the more intermediate steps that exist between primary 
activity of genes and embryologic differentiation, the larger is the target 
for environmental influences and the greater is the probability of inter- 
ference with normal processes. Ontogeny is probably associated with an 
optimal number of stepwise processes that makes the embryo a) inde- 
pendent of instability of single genetic loci and b) not overly susceptible 
to environmental modification. 

A customary experimental approach for separating roles of 2 variables 
(like heredity and environment) is to hold one variable constant. Since 
our present state of knowledge precludes precise control of environmental 
conditions for mammalian embryos I will present results based upon 
genetically standardized mammalian embryos. Identical twins would best 
suit our needs as Dr. Dunn mentioned yesterday, for with such twins we 
could study identical genotypes. Spontaneous or induced variations in 
one member of a twin would be due to environmental effects. Identical 
twinning is not available to meet our experimental requirements. How- 
ever, we do have the next best mammalian material. Here I pay homage 
to the vision and persistence of my predecessors, notably Dr. Little, who 
have contributed inbred mammalian material. For practical purposes 

Journal of the National Cancer Institute 



we have, within an inbred strain, an unlimited number of identical twins, 
i.e. genetic uniformity. 

The manner by which embryonic differentiation can be influenced by 
environment and heredity, acting both on the maternal and fetal organism, 
may be illustrated in text-figure 1. If, in our experiment, we apply two 
different environmental agents to the same genetic strain of mouse, hered- 
ity, being common to both experimental groups, can be canceled out as a 
source of variability. This experiment therefore tests for difference in 
effects of environment (A) and (B). 

Experiments on prenatal environment in which hereditary influences are canceled out 

siology < A 

Maternal physiology 


Embryonic physiology «-- 


Phenofype ( a J 


Maternal physiology 


Embryonic physiology <« 



Phenotype ( b ) 

Text-figubb 1. 

My objective to present evidence for environmental modification of 
morphogenesis will be achieved by the following steps: 1) Presentation of 
characteristics that, in genetically uniform populations, are susceptible 
to modification by environmental factors; 2) demonstration that maternal, 
experimental and spontaneous influences do in fact alter development; 
and 3) investigation of physiologic processes that produce deviation 
from the straight and narrow path of normal ontogenesis. 

A. Teratogenic Influences Applied to Different Genetic Backgrounds 

Importance of genetic background upon control of morphogenesis is 
not to be considered lightly. What happens when different genotypes are 
subjected to a common teratogenic influence? Geneticists on a number of 
occasions have performed such experiments. A single teratogenic locus 
can be introduced into different isogenic strains of mice by backcross 
breeding programs. The mutant locus may be considered the teratogenic 
agent and the recipient inbred strains will provide genetically different 
backgrounds. Often the phenotype that one gets will vary quite un- 
predictably from lethality to nonexpression, depending upon the genetic 
background within the fetus and the indirect effect of the genetic back- 
ground regulating maternal physiology. An interesting illustration of 
this is expression of the gene Fu (table 1). The genome of strain BALB/c 
and of C57BH/a permit the Fu gene to express itself in 88 and 65 percent of 
the genotypic offspring. Strain C57BE,/a genome contains normalizing 
modifiers that protect a significant number of embryos from the teratogenic 
locus. Strain C57BJR,/a mothers carrying the gene Fu exert an additional 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS op 

normalizing influence on their embryos for the mutant expresses itself in 
only 34 percent of the offspring. Thus the genetic background of the 
fetus, and in this case that of the mother, has a permissive effect on ex- 
pression of a teratogenic mutant. 

Table 1. — Percent penetrance of Fu gene in heterozygotes 




+ + 




Effect of genetic backgrounds has also been tested against experimen- 
tally administered teratogenic agents. Ingalls, Avis, Curley and Temin 
(2) subjected groups of pregnant mice from five different strains to re- 
duced atmospheric pressure. Their best example was induced "mal- 
formation" of ribs and vertebrae. Four strains treated (5 hours at 27,000 
ft. on day 9) were modified but the fifth strain of mouse was not affected. 
Although differences between strains were apparently observed, the ex- 
periment could be cited with more confidence if the authors had given a 
clearer account of the criteria by which fetuses were judged "malformed." 
Nevertheless genetic factors apparently rendered chondrogenesis in cer- 
tain strains more susceptible to modification than in other strains. 

Interaction of two genotypes with a teratogenic agent has been reported 
by Fraser and colleagues (3, 4). A given dose of cortisone (4 X 2.5 mg.) 
was administered to two strains of mice and descendents from these 
strains. Text-figure 2 has been adapted from their data for treatments 
beginning on day 11 of pregnancy. Cleft palate served as a specific 
biological end point to study inheritance of susceptibility to cortisone. 
Strains A and C57BL had susceptibilities of 100 and 18 percent, respec- 
tively. (It would be interesting to speculate why 18 percent of the 
C57BL young had cleft palate and 82 percent completely escaped.) 
Strain A females were mated with males of strain C57BL and the A 
mothers were treated as before. The anomaly, cleft palate, occurred in 
43 percent of the offspring. On first glance this might appear as a simple 
case of genetic intermediacy between the 2 parental frequencies. Hybrid 
fetuses of the same genie constitution were produced within treated 
mothers of strain C57BL. The frequency of cleft palate was 4 percent — 
even less frequent than in C57BL young gestated within resistant C57BL 
mothers. Since the two groups of hybrid young were genetically identical, 
differences in the frequencies of cleft palate (43 and 4 percent) were due 
to differences in susceptibility of the two types of mothers to cortisone 
or to differences in tr admissibility of cortisone effects to the fetus. The 
increased resistance of hybrid fetuses in C57BL mothers, as compared 
with C57BL fetuses in C57BL mothers, must be explained on the basis of 
fetal protection due to their own hybrid vigor. 

Journal of the National Cancer Institute 







□ # 



D-p© D-p-€) 

P. F 




P 2 

^fcsTRAIN C57BL 


P, 8C 


arret Fttaseit a fains tat 1931 a k alter 1934 



Text-figure 2. 

Hybrid females (A X C57BL, C57BL X A) were mated with strain A 
males. Hybrid mothers, treated with cortisone, gave an incidence of 
24 percent cleft palate, a relatively low incidence considering that the 
mother had half and fetuses had three-quarters of the genie components 
of the susceptible strain A. We may consider either partial dominance 
of resistance on the part of the fetus or hybrid vigor (delayed maternal 
influence) on the part of the mother to account for the low incidence 
of anomaly. 

Animals of another generation were backcrossed to strain A males. 
These mothers, treated with cortisone, gave offspring with 55 percent 
incidence of cleft palate. These mothers had three-quarters of strain A 
genes and the young had seven-eighths. Apparently genetic susceptibility 
had increased and heterosis decreased to permit high incidence of suscep- 
tibility to the teratogen. Expression of cleft palate was shown to be 
modifiable by the genotype expressed through maternal physiology. 
Hybrid vigor of both the mother and the fetus was able to offset, at 
least in part, the genetic susceptibility. Holding constant the teratogen 
and varying the genetic background has demonstrated the permissive 
influence of the genotype. 

Spontaneous variations within isogenic strains reared in the same animal 
house may also be regarded as reactions of different genotypes to a com- 
mon, though undefined, environment. Characterization of permissive 
effect of 5 genotypes indicates that polymorphism exists in the genetically 
uniform populations (table 2). The table reports only those anomalies 
that in each strain approximates 20 percent incidence or more. Each 
genotype has its characteristic pattern of morphologic variability. Poly- 
morphism is just as characteristic of an inbred strain of mouse as, for 
example, is its susceptibility to infectious diseases or its spontaneous 
patholigic changes that Dr. Gowen (5) and Dr. Dunn (6) reported 

Vol. 15, No. 3, December 1954 


Table 2. — Frequency percent of skeletal anomalies in 5 inbred strains of mice 







Interfrontal bone 





Parted frontal bones 



Imperfect transverse foramen #4—6 

Rudimentary ribs, vertebra #7 





Dorsal dyssymphysis, #1-20 

Perforation of neural plate, #12 or 13 

Absence of spinous process #9 




Accessory sternebrae 



Sternebral ankvlosis 



Xiphoid dyssymphysis 

Lumbar ribs, vertebra #21 



Lumbar, split centrum. . . 



Lumbar ankylosis 

Sacralization of vertebra #26 


n — 




Many of the structural variations within an inbred strain are embryo- 
logically determined as demonstrated by Green (7) and Sawin (8). Dr. 
Bailey (9) has illustrated how genetic analyses within and between strains 
can demonstrate developmental forces. A mathematical approach can 
also be a dynamic one. 

Recently we (10) restudied one of the earliest reported examples of 
nongenetic variability in the mouse. Strain C57BR was shown by Murray 
and Green in 1933 (11) to have about half of its members with ventral 
white. The size of the white patches varied from a few white hairs to 
one third of the ventral surface. Our survey, 20 years and many genera- 
tions later, has substantiated Green's observations. (Males had slightly 
more white than females but there was precious little correlation between 
the pheno types of parents and offspring.) Our method of recording white 
spots enabled us to map the distribution of white (text-fig. 3). The pattern 
provided a V-shaped area pointing posteriorly in the midline. Absence 
of pigmentation suggested a block to melanoblast migration or a wave 
of degeneration of melanoblast cells. Morphology of ribs and sternum 
in this strain of mouse (table 2) contributes additional evidence for 
similar dynamic phenomena in the lateral mesoderm. 

These examples illustrate that genetic backgrounds play a permissive 
role when environmental influences, induced or spontaneous, modify 
development. I want next to turn to the relation between pattern of 
variation of a single genotype in response to a number of environmental 
influences. What is the repertoire of a genotype? 

B. Teratogenic Agents Applied to a Single Genotype 

Responses of mice of strain 129 to a number of environmental influences 
have been studied in our laboratory (12). Spontaneous polymorphism in 
strain 129 is listed as control frequencies in column 6 of table 3. 

Journal of the National Cancer Institute 










S 30 




fij ma 

S 25 


<r 20 



| 15 



s ^y^ 



' ^^^ 

•~- IK 3 

4*r e , 


Text-figure 3. — Distribution and frequency of belly spotting in an inbred strain of 
mouse— C57BR/cd. n = 230. 

Table 3. — Frequency percent of skeletal anomalies in strain 129* 

Experimental groups 














Interfrontal bone 




66. 1 

Rudimentary ribs, vertebra 

53. 8 

Dorsal dyssymphysis #1-20 . 
Perforation of neural plate, 
#12 or 13 




28. 1 


38. 4 

Accessory sternebrae 





62. 3 

Xiphoid dyssymphysis 

29. 2 

Lumbar ankylosis, usually 
# 23 and 24 

39. 1 


23. 1 

Sacralization of #26 

Caudal ankylosis 










*When compared to controls all Items are below 1% confidence level. 

Experimental groups: 1) 0.25 mg. of trypan blue on day 7; 2) 2 hours at 25,000 ft. on day 7; 3) BALB/c mother; 
4) 0.1 mg. of cortisone on day 7; 5) litters 3-7. 

Five other samples of this strain were surveyed after mothers and em- 
bryos were subjected to various treatments, viz., 0.25 mg. of trypan blue 
on day 7, reduced atmosphere of 25,000 feet on day 7, gestation in BALB/c 
mothers, 0.1 mg. of cortisone on day 7, and gestation in elderly parents. 
Frequencies of anomalies are given in the table only in instances that are 
significantly different from the control sample. Two points can be drawn 
from the data, a) Those structures that vary in the controls, i.e., suscep- 
tible characters, were altered by the treatments employed, b) Just as 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 

did each genotype have its characteristic polymorphism, so did each 
treatment have its characteristic pattern of anomalies. 

One of the susceptible characters in strain 129 has been selected for 
presentation in more detail. It is more than coincidence that I selected 
the trait that Dr. Green (7) has so skillfully analyzed. Differentiation of 
vertebra #26 in mice of strain 129 finds itself in an ontogenetically uncer- 
tain situation. It can become a) a typical lumbar as happens in most 
strains of mice, b) a typical sacral or c) something in-between (fig. 1). 
Strain 129 is particularly well suited for our study because the three types 
of vertebrae occur with about equal frequency. The sacralization influ- 
ence of all treatments, so far investigated, are listed (table 3, text-fig. 4). 
The data show a general tendency for treatments on day 7, improvement 
of maternal diet, and gestation in BALB/c mothers to reduce the frequency 
of sacralization. Offspring from young parents were not significantly 
altered but offspring from older parents showed a tendency to increased 
incidence of sacralization. Inherited propensities in strain 129 were not 
necessarily final judgments for the embryo. The probability of occurrence 
of anomaly could be shifted so that one could reduce or increase the 
incidence of sacralization. Thus the environmentally susceptible char- 
acter could be enhanced, suppressed, or not affected depending upon the 
environmental agent. 

Environmental influences applied to the embryo, so far presented, have 
been applied empirically. Variations observed have not indicated path- 
ways by which environment modifies morphogenesis. After learning what 
agents can alter development, an inquiring mind wants to know how a 
given agent can create embryonic imbalance. Since the agents so far 
reported did not provide clues about mode of action we have used another 

0.25 ntq TRYPAN 
DAY 7 

85.000 FEET 








I AND 2 







46 <0.0I 

54 <0 .05 

56 <0.05 

136 <0.0l 

74 <0.05 

150 >0.05 

110 <0.0l 

O 10 tO 10 40 SO 60 70 80 

Text-figure 4. — Effects of environmental factors upon sacralization of vertebra #26. 

Journal of the National Cancer Institute 



teratogen on strain 129. Experiments are still in progress and interpreta- 
tions must be tentative until more of the story has unfolded. 

Experiment 1. — Mice of strain 129 were fasted for 24 hours during the 
9th day of pregnancy (5 to 20 somite period of the embryo). Anomalies 
illustrated in figures 2 and 3 occurred in 32 percent of the fetuses. Using 
these defects, pseudencephaly and fused ribs, as end points, we have been 
investigating how fasting results in anomaly (table 4) . 

Table 4 




of fetuses 



Fast — 24 hours 






Folic-acid-deficient diet ad libitum 


Folic-acid-deficient diet containing X-methyl 
folic acid 


4 and 5 


CHO (0.8 gm. of glucose or 0.1 gm. of casein) . . . 
Fast — 24 hours force-fed 2 cc. H2O 


Experiments 2 and 3. — The types of congenital defects observed and 
the short effective period of treatment suggested similarity to effects of 
folic-acid depletion in rats. A semisynthetic diet, deficient in folic acid, 
fed during the 9th day of pregnancy completely protected young from the 
effects of fasting. Furthermore, folic-acid antagonist administered on 
day 9 to fasting mothers did not appreciably augment the degree or fre- 
quency of anomaly. These observations precluded folic-acid depletion 
as a causative agent for the teratogenic effect of fasting. 

Experiments 4 and 5.- — The semisynthetic diet that protected embryos 
from anomaly contained more than a dozen ingredients. We were faced 
with the likelihood of 11 additional experiments to eliminate one item 
each time. Instead, we reversed the approach and fed by stomach tube 
the mothers that were otherwise fasted. One group received a solution 
of vitamin-free casein to offset possible amino acid depletion and another 
group received glucose solution by mouth to ward off energy depletion. 
Both treatments protected the embryos, so we have reasoned (a) that 
protein and glucose had a sparing effect on embryonic tissue protein or 
(6) that protein and glucose had protected by preventing depletion of 
energy rich carbohydrate in the form of glucose directly or by conversion 
from amino acids. 

Experiment 6. — There exists precedent, from the work of Landauer (IS) 
and Zwilling (14) with the chick embryo, to suspect that interruption of 
carbohydrate metabolism could induce anomaly. Circulating blood 
glucose was measured at the end of 24 hours of fasting. Sure enough, 
the glucose level was down 75 percent. The hypoglycemia, protection 
by glucose, and analogy with the chick provided a working hypothesis to 
explain the cause of anomaly. 

Experiment 7. — We turned our attention to the embryos to correlate 
morphogenesis with the time of induction of anomalies. We found at 

Vol. 15, No. 3, December 1954 

646 proceedings: symposium on 25 years of 

the end of day 10 that a proportion of embryos showed retention of open 
brain tubes. Pseudencephaly induced by fasting was a result of failure 
of the closure of the neural tube. 

Experiment 8. — Piecing together odd bits of observation we recalled 
that fasting mice are said not to drink. Mice did appear shrunken after 
fasting 24 hours. Furthermore, hadn't we protected embryos by giving 
glucose and protein in solution? Fasting animals were given isotonic 
saline on the same schedule used for glucose protection. Anomalies were 
observed in 26 percent of the fetuses thereby showing that the water in 
which the protein and glucose was suspended did not provide protection. 
Altered carbohydrate metabolism as a teratogenic influence is a tempting 
hypothesis at the moment and is being investigated further. 


Inbred mammalian material has offered opportunity for genetic stand- 
ardization so that environmental influences on prenatal life could be in- 
vestigated. By canceling out genetic variability one learns that environ- 
mental factors influence morphogenesis in the embryo. 

Geneticists in the first half of the current century have firmly established 
basic Mendelian inheritance. I would suggest that Mendelian principles 
have been derived from carefully selected samples of deviants. Most 
variations in man and his domestic animals require bulky mathematical 
extensions of the simple Mendelian situations. Variations in normal 
development are most likely inherited by what Dr. E. L. Green (7) has 
termed multiple-factor situations with thresholds. 

For the sake of clarity I would like to point out that our objectives 
should be clearly distinguished from those commonly pursued in efforts 
to study effects of genie mutants in embryology. I completely divorce 
myself from the philosophy that a study of abnormal development will 
explain its counterpart in normal physiological embryology. Nor do we 
wish to search for earliest visible effect of any given gene. 

"Skepticism should be applied to the hope that studies on the develop- 
mental effects of major mutant genes will lead to an understanding of the 
role in growth and development of those genes which make up the normal 
genotype of fowl or any other higher organism. The damage done by a 
mutation may, and often does, find expression in the abolishment of a 
particular and well-defined metabolic function. It does not follow, though 
this assumption is commonly made, that the normal allele is the sole or 
even the major factor in this particular and well-defined function (13)." 

Teratogenesis is a function of background inheritance. Genetically 
induced and intangible embryonic imbalances seem to interact with per- 
missive influence of background inheritance to produce congenital de- 
formity. Teratogenesis is also a function of environmental factors. 
Different teratogenic agents (or doses) bring out characteristic patterns 
of polymorphism. Mild environmental agents tend to affect those char- 
acters that are spontaneously variable. Anomalies on a given genetic 
background can be enhanced, suppressed or not affected. Careful appli- 

Journal of the National Cancer Institute 



cation of teratogenic agents may yield clues about metabolic pathways 
that lead to deformity in mammals. 


(1) Gruneberg, H.: Embryology of mammalian genes. Revue Swisse de Zoologie 

57: 129-139, 1950. 

(2) Ingalls, T. H., Avis, F. R., Curley, F. J., and Temin, H. M.: Genetic deter- 

minants of hypoxia-induced congenital anomalies. J. Hered. 44: 185-194, 

(3) Fraser, F. C, and Fainstat, T. D.: Production of congenital defects in the 

offspring of pregnant mice treated with cortisone. Pediatr. 8: 527-533, 1951. 

(4) Kalter, H.: The inheritance of susceptibility to the teratogenic action of corti- 

sone in mice. Genetics 39: 185-196, 1954. 
^5) Go wen, J. W.: Significance and utilization of animal individuality in disease 
research. J. Nat. Cancer Inst. 15: 555-570, 1954. 

(6) Dunn, T.: The importance of differences in morphology in inbred strains. J. 

Nat. Cancer Inst. 15: 573-589, 1954. 

(7) Green, E. L.: Quantitative genetics of sketal variations in the mouse. I. Crosses 

between three short-ear strains (P, NB, SEC/2). J. Nat. Cancer Inst. 15: 
609-654, 1954. 

(8) Sawin, P.: Discussion following (7). J. Nat. Cancer Inst. 15: 625-627, 1954. 

(9) Bailey, D.: Discussion to inheritance of susceptibility to congenital deformity — 

embryonic instability. J. Nat. Cancer Inst. 15: 651, 1954. 

(10) Runner, M. N., and Morris, N.: Unpublished. 

(11) Murray, J. M., and Green, C. V.: Inheritance of ventral spotting in mice. 

Genetics 18: 481-486, 1933. 

(12) Runner, M. N.: Modified differentiations of the sacrum. (Abstract.) Anat. 

Rec. 115: 364, 1953. 
(18) Landauer, W.: The genetic control of normal development in the chicken 

embryo. Ann. New York Acad. Sc. 55: 172-176, 1952. 
(14) Zwilling, E.: The effects of some hormones on development. Ann. New York 

Acad. Sc. 55: 196-202, 1952. 

Vol. IS, No. 3, December 1954 


proceedings: symposium 

Plate 45 

Figure 1. — Differentiations of vertebra #26 in strain 129. On the left is a typical 
lumbar, on the right a typical sacral, and in the middle an asymmetrical vertebra. 

Figure 2. — Pseudencephaly in mice of strain 129 following treatment (fasting) during 
the ninth day of pregnancy. On the right is a normal sibling. 

Figure 3. — Rib fusions in mice of strain 129 following treatment (fasting) during the 
ninth day of pregnancy. These young were taken from the mother at 19 days 
postcoitum. On the right is a normal sibling. 



2 26 

t* On 

> 28 
j 29 

W * V 



316263—54 25 

Dr. Donald W. Bailey, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

The outmoded question of which is the more important, heredity or environment, 
has been replaced by the more constructive and reasonable question of how do environ- 
ment and heredity work together to produce the well-organized developmental patterns 
which we observe. The work that Dr. Runner has just presented is an approach 
toward answering this question. Just where do the environmental and genetic path- 
ways meet during the course of development? What kind of profile of gene action 
can be obtained by these environmental probings? In my opinion, this approach 
should prove to be quite fruitful. 

I would like to supplement Dr. Runner's work on the relationship of environment 
to heredity with a preliminary polygenic study on how the genes and environments 
differ in their affects on the shape of a vertebra. 

In this study the second cervical vertebra was measured in most of its dimensions 
as seen in posterior view. From these measurements it was possible to obtain correla- 
tion coefficients for each combination of measurements. Thus it was possible to obtain 
a complete set of such correlations. 

In fact, by the use of the methods of analyses of variance and covariance it was 
possible to separate such a set of correlations into three kinds, according to the source 
of control : Genetic, inter-maternal environmental, and intra-maternal environmental. 
Thus we have three complete sets of coefficients showing the relationships of the 
dimensions to each other. 

Now, the central question involved in this investigation was: If these sets of 
coefficients differ, one from another, then in what way? Or in other words, do the 
genes and the two kinds of environments differ in their affects on the shape of the 

In order to compare these three groups of relationships, we should formulate a way 
to compare all dimensions of the vertebra at once and their relative changes. The 
method used was as follows: If one dimension, arbitrarily chosen, is increased in 
size, what are the correlated responses in the other dimensions? For instance, if 
we should increase the width of the neural arch, what changes would occur in the 
other dimensions? And, are these responses different for the genetic and environ- 
mental sources of control? 

Conclusions derived from alternately increasing each dimension and noting responses : 

Genetic: vertebra was not generally altered in all dimensions at once; sometimes a 
negative response was observed, the two arch dimensions were found to be independent, 
but the centrum dimensions were dependent. 

Inter-maternal environment: there was a very strong tendency to alter the angle of 
the transverse process, arch dimensions were independent, and centrum dimensions 

Intra-maternal environment: the dimensions were generally all altered excepting the 
centrum dimensions which were independent. 

This study has shown that the two kinds of environments and the genes apparently 
differ in their affects on vertebral shape. 

The findings of this study are quite in harmony with those of Dr. Runner's. He has 
shown how specific alterations in the environment can show relatively specific affects 
upon development. This specificity of environmental action is perhaps an acceptable 
explanation of the differences in the environmental and genetic relationships in the 
polygenic study, i.e., the environmental change acts disproportionally upon different 
portions of the genie system determining the developmental pattern. I think we should 
expect areas of vulnerability in a genie system to a specific environmental stress — both 
in time and substance. If this is the case, then these environmental probings of 
Dr. Runner's should, in time, provide a well-defined profile of gene action in develop- 
ment and perhaps eventually a knowledge of primary gene action itself. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Session IV. Genetic Control of Func- 

Chairman, Dr. Harrison R. Hunt, Professor 
Emeritus oj Zoology, Michigan State College, East 
Lansing, Mich. 

Speaker: Dr. Herman B. Chase 
Some Examples oj Gene-Controlled Functional Disturbances in the Mouse 
Discusser: Dr. Elizabeth S. Russell 

Speaker: Dr. George D. Snell 

The Enhancing Effect (or Actively Acquired Tolerance) and the Histo- 
compatibility^ Locus in the Mouse 
Discusser: Dr. N. A. Mitchison 

Speaker: Miss Margaret Dickie 

The Expanding Knowledge oj the Genome oj the Mouse 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Introduction: Session IV 
Dr. Harrison R. Hunt, Chairman 

Genetics is an aspect of physiology. It is concerned with the details of the processes 
by which an organism acquires its adult form and functions. The accomplishments of 
a gene are determined not only by its own activities, but by the environment in which 
it operates, and this environment comprises other genes, agencies outside the organism 
operating upon it, and the status of the organism at the time the gene becomes func- 
tional. Embryology and experimental morphology are the morphologic aspects of 
this developmental process. If the end result of such gene-influenced development is 
definitely harmful, one is concerned with pathology, as in the appearance of a malig- 
nant neoplasm. 

Rats, mice, and men are good subjects for such investigations in physiological genetics. 
For example, Heston's work on the bent-nosed rat, Jay's and Burrington's on the 
flexed-tailed mouse, and Hunt's and Hoppert's studies on the causes of dental caries 
in rats have shown that both the diet and the genes are responsible for the phenotype. 
It is suggested that some adaptation of Wright's method of path coefficients might be 
useful in estimating the relative importance of heredity and various environmental 
agencies in at least some cases. Studies like those of Dr. Chase and Dr. Snell, using 
rodents, should be undertaken with numerous traits. The rapidly growing list of 
genes in the mouse, listed by Miss Dickie, is encouraging, because every new gene 
discovered and located on a chromosome is a potential tool in investigations of the 
genetic control of function. 

Some Examples of Gene- Controlled 
Functional Disturbances in the 
Mouse u 2 

Heeman B. Chase, Biology Department, Brown 
University, Providence, R. I. 

In the study of physiological genetics, it becomes evident that there are 
no clear dividing lines that separate morphology, physiology, and behavior. 
A certain morphologic character is the result of certain gene-modified 
physiological processes. A certain behavior or function is in turn the 
result of certain morphologic characteristics. The morphologic charac- 
teristic in question might be an enzyme, a hormone, the number of types 
of cells, or some structure of the skeleton or nervous system. As employed 
in this paper, then, a functional disturbance consists of the description of 
some step in the gene-controlled process which leads to a morphologic 
variation or the description of some function resulting from a morphologic 
variant. Schematically this is shown in text-figure 1. Five rather 
varied examples from my experiences with mice are presented here. 

Insulin tolerance. — There is a strain of mouse that tolerates 200 units of 
insulin at 35 days of age, instead of the normal one unit (1, 2). This 
characteristic is not due to a single gene difference but probably to about 
three. In recent experiments we find that this mouse has an extremely 
high level of "insulinase" or some such substance in the liver. With an 
injection of insulin the blood-sugar level of this mouse begins to drop as 
in the case of the normal but levels off and rises again instead of continuing 
to the critical 15-mg. percent level. In vitro studies, involving extracts 
of liver homogenate from these tolerant mice, indicate the presence of 
"msulinase" in quantities sufficient to counteract 800 or more units of 
insulin per mouse liver in 30 minutes (3). The presence of an insulinase- 
inhibitor in intact animals is indicated. These insulin- tolerant mice are 
also more resistant to alloxan-induced diabetes (4). The study is not 
completed, but it is possible that some substance may be present which 
can inhibit or destroy alloxan and thus protect islet cells of the pancreas 
from damage. 

Anophthalmia. — This strain of eyeless mice was obtained originally 
from Dr. C. C. Little of the Jackson Memorial Laboratory. Anophthal- 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 28, 1954. 

2 Some of the work summarized here was supported, in part, by grants from the U. S. Public Health Service 
and by grants from the American Cancer Society recommended by the Committee on Growth of the National 
Research Council. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 
816263—54 26 


proceedings: SYMPOSIUM ON 25 YEARS of 




Specific cellular 
enzyme or antigen 

Differentiated cell 

Primary gene action 

biochemical processes 


Ccellular or extra- cellular) 

Cdiffusible substances, 

hormones, granules, etc.) 

"Gross variant" 

roduction of specific 

action of substances 


Iross function or 

Text-figure 1. — Each morphologic entity is preceded and followed by a physiological 
process. Likewise, each function can be viewed as being bracketed by two levels of 

mia is a single factor recessive in some crosses, whereas two major factors 
are involved in other crosses (5-8) . A variety of pleio tropic effects stems 
from the failure of the optic vesicle to form a normal cup and to make 
contact with the ectoderm to induce a lens. Among the later effects 
are the failure to develop of certain layers of the superior colliculus and 
of the visual cortex (lack of induction due to absence of the optic nerve), 
and the atrophy of the pioneer fibers of the trochlear and abducens nerves, 
following atrophy of the extrinsic eye muscles due to absence of the eye- 
ball. Normally, the optic vesicle when about 100 /x in diameter forms a 
cup and starts to induce a lens. In the anophthalmic embryos, this 
vesicle forms no cup or forms several small abortive cups and thus fails in 
its initial function of inducing a lens and thereby an eyeball, optic nerve 
fibers, etc. 

Pigment cells. — A few (of the order of 6 or less in zigzag follicles) den- 
dritic pigment cells in the hair bulb supply pigment granules to cells of the 
upper bulb which are to become the cells of the medulla and of the cortex 

Journal of the National Cancer Institute 



(9-18) . Some granules remain outside the cells and are thus carried along 
as the hair grows, but many granules are actually taken up by the epi- 
thelial cells of the bulb. Depending on the genotype, the epithelial cells 
then modify the laying down of additional melanin on these granule sites. 
Some genes which affect the dendritic pigment cells themselves and thereby 
their function are: the gene for albinism (c) that causes a failure of mel- 
anin formation even though the cells are present; the gene for white spot- 
ting (s) that causes a failure of these cells in whole areas; the gene for 
silver (si) that causes a failure of these cells in scattered hair follicles; and 
the gene for dilution (d) that causes a few of these cells to have large 
clumps of pigment rather than the usual very small granules. These 
clumps generally do not become incorporated into the epithelial cells but 
rather result in large masses of pigment along the hair shaft, so massive 
that the shaft is often distorted at these points. The case of silver is of 
particular interest because it can be simulated so precisely by X-ray de- 
struction of pigment cells, the X-ray graying effect. With appropriate 
modifiers the gene for silver results in a phenotype which we have called 
basal dilution. In this condition there are some all-white and mosaic 
hairs as in the ordinary silver, but practically all the remaining hairs have 
white or diluted bases. The explanation for this latter condition is not 
yet fully determined but it appears possible that in addition to there being 
fewer pigment cells there is also a shortening of the period of melanogenic 
activity for these cells. 

Except for skin on the feet, tail, and ears, there are normally no actively 
melanogenic pigment cells in the epidermis of the adult mouse. Such 
cells are present during the first few days after birth (13), however, and 
can be made melanogenic again by irritants. They also become mildly 
active in old hairless (hr/hr) mice. 

Yellow obesity. — The obesity associated with the yellow gene (A v ) is 
generally lost following inbreeding. With a high fat diet, however, such 
inbred mice quickly become obese, whereas their nonyellow litter mates 
do not. The function of fat metabolism is thus altered by the presence of 
the A v gene (14, 15). The difference is not due entirely to the amount of 
food consumed. 

Hairlessness. — The skin functions in a precisely integrated cyclic man- 
ner (11, 16). The corium and adipose layer become thicker with the ana- 
gen phase of the hair growth cycle of the hair. The surface epidermis by 
increased mitotic activity becomes thicker in early anagen; but in later 
anagen, when the hair shaft is being formed, the epidermis becomes even 
thinner than during the resting phase of hair growth, telogen. There are 
simple feed-back mechanisms of control, as for the amount of corneum and 
sebum, and in addition, there are controls, possibly of a competitive nature, 
imposed by the great activity of the hair bulb. In this system of cyclic 
and interacting processes, a modification of one process is likely to have 
an indirect effect on many processes. One such condition is that of gene- 
controlled hairlessness. 

Vol. 15, No. 3, December 1954 

658 proceedings: symposium on 25 years of 

Animals which are homozygous for this gene (hr/hr) grow a normal first 
coat but at the first catagen stage an abnormal club is formed and the fol- 
licle fails to shorten for a normal telogen follicle (11, 17, 16, 13, 18). The 
hair falls out, having no normal club, and the follicle strand becomes at- 
tenuated and breaks into isolated portions. These isolated fragments 
develop into sebaceous and later keratinized cysts. The fragment which 
includes the dermal papilla can later produce an abortive hair in a cyst, but 
the continuity of the follicle with the surface epidermis is lost. In normal 
catagen the connective-tissue sheath around the epithelial external sheath 
forms a thick glassy membrane that wrinkles and shortens as the lower 
follicle degenerates and thus holds the remaining cells and the dermal 
papilla in contact with the permanent upper portion of the follicle. The 
pressure by the glassy membrane probably also induces the brushlike 
club end of the hair. In the hairless mouse the connective-tissue sheath 
fails to produce a glassy membrane and thus to perform one of its normal 
functions (figs. 1 and 2). 

As a result of the hairless condition, these mice have a greater food con- 
sumption and activity than have the heterozygous (Hr/hr) haired mice of 
the same strain. This difference is less pronounced when the animals are 
kept at a temperature of 78° F. than when they are kept at about 70° F. 
Furthermore, the epidermis and associated corneum become thicker with 
each ' 'cycle" and fail to return to the normal original thickness. The 
cysts continue to grow and almost completely fill the corium. The adi- 
pose layer becomes increasingly thin. In effect, the whole skin "ages" 
more rapidly than would be the case if functional hair follicles were still 
present. Also it might be pointed out that the skin of young hairless 
mice develops sebaceous adenomas very readily, especially with the 
application of methylcholanthrene. 

Conclusions. — These five cases represent rather diversified examples of 
gene-modified functions. They range from an ability to restore the blood- 
sugar level, in spite of large injections of insulin, to the action of pigment 
cells in producing coat-color characteristics; from a change in fat metab- 
olism to the inductive action of the optic vesicle and cup; from the action 
of the connective-tissue sheath in the hair growth cycle and indirectly in 
determining a thermal optimum to a mechanism that resists alloxan- 
induced diabetes. Mechanisms which control embryonic development, 
which influence cyclic activities, and which help maintain homeostasis in 
the organism are here represented. For such studies the mouse is an 
excellent organism. 


(1) Chase, H. B. : Inheritance and selection of insulin-resistance in mice. (Abstract.) 
Genetics 35: 101, 1950. 

(#) Chase, H. B., Gunther, M. S., Miller, J., and Wolffson, D.: High insulin 
tolerance in an inbred strain of mice. Science 107: 297-299, 1948. 

(8) Beyer, R. E. : A study of insulin metabolism in an insulin tolerant strain of mice. 
Thesis, Brown Univ., 1954. 

(4) Frankenhuis, B.: Effects of alloxan on an insulin tolerant strain of mice. The- 
sis, Brown Univ., 1953. 

Journal of the National Cancer Institute 


(5) Chase, H. B.: Studies on an anophthalmic strain of mice. III. Results of 

crosses with other strains. Genetics 27: 339-348, 1942. 

(6) : Studies on an anophthalmic strain of mice. IV. A second major gene 

for anophthalmia. Genetics 29: 264-269, 1944. 

(7) : Studies on an anophthalmic strain of mice. V. Associated cranial nerves 

and brain centers. J. Comp. Neurol. 83: 121-139, 1945. 

(8) Chase, H. B., and Chase, E. B.: Studies on an anophthalmic strain of mice. I. 

Embryology of the eye region. J. Morphol. 68: 279-301, 1941. 

(9) Chase, H. B.: Greying of hair. I. Effects produced by single doses of X-rays on 

mice. J. Morphol. 84: 57-80, 1949. 

(10) : Number of entities inactivated by X-rays in greying of hair. Science 

113: 714-716, 1951. 

(11) : Growth of the hair. Physiol. Rev. 34: 113-126, 1954. 

(12) Chase, H. B., and Rauch, H.: Greying of hair. II. Response of individual hairs 

in mice to variations in X-radiation. J. Morphol. 87: 381-391, 1950. 
(18) Chase, H. B., Rauch, H., and Smith, V. W.: Critical stages of hair development 
and pigmentation in the mouse. Physiol. Zool. 24: 1-8, 1951. 

(14) Chase, H. B., and Fenton, P. F.: The expression of obesity in yellow mice. 

(Abstract). Genetics 36: 546-547, 1951. 

(15) Fenton, P. F., and Chase, H. B.: Effect of diet on obesity of yellow mice in 

inbred lines. Proc. Soc. Exper. Biol. & Med. 77: 420-422, 1951. 

(16) Chase, H. B., Montagna, W., and Malone, J. D.: Changes in the skin in rela- 

tion to the hair growth cycle. Anat. Rec. 116: 75-81, 1953. 

(17) Chase, H. B., and Montagna, W.: The development and consequences of hair- 

lessness in the mouse. Genetics 37: 573, 1952. 

(18) Montagna, W., Chase, H. B., and Melaragno, H. P.: The skin of hairless 

mice. I. The formation of cysts and the distribution of lipids. J. Invest. 
Dermat. 19: 83-94, 1952. 

Vol. 15, No. 3, December 1954 

660 proceedings: symposium 

Plate 46 

Figure 1. — Normal catagen hair follicle with normal club and thickened connective 
tissue around lower follicle. X 485 

Figure 2. — Hairless catagen hair follicle with abnormal club and no thickened con- 
nective-tissue sheath around lower follicle. X 485 





316263—54 27 

Dr. Elizabeth S. Russell, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

During the past half hour I have been alternately enjoying Dr. Chase's excellent pres- 
entation, and worrying for fear he would erase a most intriguing figure which has given 
me an inspiration for this discussion. That figure shows a line across the board going 
from morphology to physiology and back again; from gene, to primary gene action, to 
cellular enzyme, to biochemical process, to differentiated cell, and so on all the way out 
to the final possible stage of extra-organismic character. This line represents to me not 
only an excellent picture of paths of gene-action ; it is also a very clear picture of a zigzag 
hair I It is in the study of hair growth and pigmentation that Dr. Chase's researches 
and my own have most in common. Actually, our investigations have tended to 
complement each other in this field. When I attempt to study the pigment granules 
in a certain hair, and deduce from them something of the story of gene-controlled pig- 
ment deposition, my first question is always, "Just where are we along the hair shaft? 
This may mean something very different at the tip from what it would near the base." 
Physiological genetics investigation is very much like that zigzag hair, and it is very 
important to attempt to trace gene actions all the way in both directions, from the first 
effect within single cells to the last consequence in organismic behavior. It is for this 
reason that "major" genes, with effects easily recognized at many different stages, have 
been extensively and intensively studied by physiological geneticists. They can make 
use of both qualitative and quantitative differences associated with genie substitutions, 
and can study both multiple pleiotropic effects of single-gene substitutions and changes 
in single characters brought about by the actions of large numbers of genes not in- 
dividually identifiable. As I have been involved in studies dealing with both ends of 
this spectrum, I would like to quote an example from each extreme. 

Major effects produced by single-gene substitutions may be exemplified by studies 
of the anemia-producing effects of the deleterious alleles of the TF-series in the mouse. 
These effects appear to be mediated through an arrest in the maturation of hematopoi- 
etic cells in the bone-marrow (1). It has further been shown by isotope incorporation 
experiments that this arrest involves a great delay in formation of one portion only of 
the hemoglobin molecule (2). Protoporphyrin, and consequently heme, is formed 
much more slowly in the anemics than in the normals, while the globin portion of the 
molecule is formed at essentially the same rate in both. This seems to me to be 
approaching the level of analysis of primary gene action; at least we have gotten inside 
the cell. However, a word of caution should be inserted here; it seems rather probable 
to those of us who have been involved in these studies that the difference in rate of 
protoporphyrin formation may reflect rather than cause the delayed maturation, and 
this interpretation puts our observations much further from primary gene action than 
would be true if the lesion in prophyrin formation caused the arrest. 

At the other end of the scale, the investigations in the Inbred Nucleus of the Jackson 
Laboratory have disclosed a number of physiological differences known to be partially 
genetically determined because they differ much more among than within inbred strains 
of mice. Usually it is not possible to attribute such differences to particular identifiable 
genes, but there is no doubt of the genetic nature of strain differences. More and more 
such differences are coming to be recognized. In the library of the Jackson Laboratory 
there are more than 1,200 references from journals (1951-53) alone to studies based on 
inbred mice. These have been catalogued and classified in a subject-strain bibli- 
ography (8) which provides rapid access to literature pertinent to a particular strain 
or condition. From this list I have gleaned a variety of newly recognized physiological 
strain differences (4) > These include differences in the nature of disease produced by 
a certain pathogen, in the survival time of infected individuals, and in susceptibility 
and resistance. Strains differ in capacity for antibody production, and in reaction to 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

664 proceedings: symposium 

changes in environment, for example, cold tolerance and sensitivity to toxic effects of 
chloroform. Strains differ in content of particular hormones, sensitivity to particular 
hormones, type of endocrine balance, and in reaction to extirpation of endocrine organs. 
In the Inbred Nucleus we have particularly studied variations in blood-cell characters 
(6, 6), variations in longevity, and differences in types of pathology characterizing 
different inbred strains (7) . 

We are coming into a new stage in genetics, where an ever-increasing number of 
inherited physiological differences will be recognized. I hope these new ones may be 
analyzed with skill and competence such as Dr. Chase has beautifully demonstrated 
in his excellent presentation this evening. 

Literature Cited 

(1) Russell, E. S., Snow, C. M., Murray, L. M., and Cormier, J. P.: The bone 
marrow in inherited macrocytic anemia in the house mouse. Acta Haemat. 10: 
247-259, 1953. 

(#) Russell, E. S., Altman, K. I., Salomon, K., and Scott, J. K.: Chemical charac- 
terization of the erythropoietic defect in T^TF* anemic mice. (Abstract.) Ge- 
netics 38: 687, 1953. 

(3) Staats, J.: A classified bibliography of inbred strains of mice. Science 119: 

295-296, 1954. 

(4) Russell, E. S.: Significance of the physiological pattern of animal strains in bio- 

logical research. Proc. 4th Ann. Meet., Animal Care Panel, 1953 pp. 140-150. 

(5) Russell, E. S., Neufeld, E. F., and Higgins, C. T.: Comparison of normal blood 

picture of young adults from 18 inbred mouse strains. Poc. Soc. Exper. Biol. & 
Med. 78: 761-766,1951. 

(6) Budds, 0. C, Russell, E. S., and Abrams, G. E.: Effects of genetics and anes- 

thesia upon granulocyte and agranulocyte levels in seven inbred mouse strains. 
Proc. Soc. Exper. Biol. & Med. 84: 176-178, 1953. 

(7) Russell, E. S., Fekete, E. A., Borges, P. R. F., MacFarland, E. K., and 

Collins, J. B.: Longevity and pathology patterns of mice from ten major 
inbred strains. Anat. Rec. 117: 547-548, 1953. 

The Enhancing Effect (or Actively Ac- 
quired Tolerance) and the Histocom- 
patibility-^ Locus in the Mouse u 2 

George D. Snell, 3 Roscoe B. Jackson Memorial 
Laboratory, Bar Harbor, Maine 

At an anniversary gathering such as this, more than at an ordinary 
scientific meeting, we allow ourselves the privilege of recalling the past 
and questioning the future. In keeping with this tradition, I am going 
to start my remarks with a fragment of history and conclude them with 
a dash of speculation. In between these boundaries I shall present what 
I hope is a fairly substantial body of experimentally verified fact. 


The history begins in 1914 at the Bussey Institution. In that year 
Dr. Little, then just completing his graduate studies under Dr. Castle, 
published a paper in Science (1) which brilliantly anticipated the now 
generally accepted genetic theory of transplantation. For some months 
previously Dr. Little had been spending part of his time in Tyzzer's labora- 
tory at the Harvard Medical School, making crosses between partly inbred 
strains of mice, and inoculating the various hybrid generations with 
transplantable tumors. Definite ratios of susceptibility and resistance 
were obtained, but they simply did not conform to any then recognized 
as likely to result from segregating Mendelian factors. Susceptibility 
was manifest in the Fi in the manner characteristic of dominants, but 
appeared in an unexpectedly small proportion of the F 2 and backcross 
hybrids. The 1914 paper, while not mentioning the transplantation 
experiments, showed that ratios of just this sort were to be expected in 
the case of characters due to the interaction of multiple dominant factors. 

Dr. Little maintained his interest in transplantable tumors after his 
departure from the Bussey, and a program for the study of the genetics 
of transplantation which was initiated at the University of Michigan, and 
continued at the Jackson Memorial Laboratory after its founding, led to 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 28, 1954. 

3 This investigation was supported by a research grant C-1329(C2) from the National Cancer Institute of the 
National Institutes of Health, U.S. Public Health Service; by a grant-in-aid from the American Cancer Society 
upon recommendation of the Committee on Growth of the National Research Council; and by a grant to the 
Roscoe B. Jackson Memorial Laboratory from the Anna Fuller Fund. 

3 Most of the experiments here reported were carried out by Miss Priscilla Smith, to whom it is a pleasure to ex- 
press my deep appreciation. Other data are taken from published and unpublished experiments of Dr. Kaliss 
to whom I am indebted for permission to quote them. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

666 proceedings: symposium on 25 years of 

our present understanding of the essentially genetic basis of transplant 
susceptibility and resistance [see reviews by Little (2) and Snell (8)]. The 
work that I am going to describe is a natural outgrowth of these early 

The Experimental Background 

The study of the mechanisms which determine susceptibility and 
resistance to tumors and normal tissue transplants has now become the 
concern of several members of the laboratory staff, who share a common 
interest in this field. Of the various methods of approach now being 
used I shall be largely concerned with only two. All phases of this pro- 
gram, however, form an interlocking whole, and in many respects I am 
indebted to research assistants, visiting fellows, and staff members, and 
most certainly to Dr. Little himself, for stimulation and help that I cannot 
acknowledge in detail. 

The first subject which I wish to discuss is a phenomenon concerned 
with the abrogation of the resistance of the host to tumor homografts 
{i.e., grafts between strains). This phenomenon has variously been called 
the "XYZ effect," the "enhancing effect/ ' and " conditioning the host" 
(4~7). The names " immunological paralysis" and "actively acquired 
tolerance" have been applied to phenomena, investigated in other labora- 
tories, which involve different experimental procedures but probably the 
same fundamental biological mechanism (8-10). 

As investigated in this laboratory, the typical procedure for producing 
the enhancing effect involves the pretreatment of the host with prepara- 
tions of normal or tumor tissue taken from the same strain as the graft 
donor. The method of preparing the extract first employed by us (11), 
and still one of the most satisfactory, is to lyophilize the tumor tissue and 
then to suspend the resulting dry powder in distilled water or saline prior 
to injection. Injections, usually 2 to 10 in number and totaling usually 
5 to 50 mg. of dry tissue, are given intra-abdominally, and the living tumor 
is grafted subcutaneously about a week after the last injection. 

With some, but not all, tumor-host combinations, most of the injected 
or experimental animals succumb to progressively growing tumors, while 
the controls with rare exceptions survive. The incidence of death in 
experimental groups is frequently 80 to 100 percent. 

Several years of investigation by the Jackson Laboratory group have 
revealed a number of important characteristics of the enhancing effect. 

Kaliss and Day (12) have investigated the time relations of the pheno- 
menon. The injections of tissue preparation are ineffective if given 
more than one day after the tumor graft, and relatively ineffective at one 
day. The effect is clearly evident, however, if the injections precede the 
graft by 1 day only. In evaluating these relationships it must be re- 
membered that it takes a tumor implant several days to become estab- 
lished — it is usually a week before there is a definitely palpable mass — so 
that the effective interval may be longer than the experimental procedure 
would imply. 

Journal of the National Cancer Institute 


The effect persists. There is no diminution in tumor growth when the 
graft is made 19 weeks after the last injection, and little if any at 40 weeks. 

The effect can be passively transferred (13, lJj). Serum from injected 
mice will induce graft susceptibility in otherwise untreated mice of the 
same host strain. Active serum can also be obtained from rabbits injected 
with appropriate preparations of mouse tissue. Further studies are in 
progress (Kaliss). 

The effect can be produced with preparations of certain normal mouse 
tissue (15). Of several tissues tested, spleen is apparently the most active 
(at least in the tumor-host combination usually employed), kidney the 
next, liver the least. Washed red cells of mice produce no effect. 

The effect is species-specific (16) . Tissue from rats, hamsters and guinea 
pigs produce no enhancement of the growth of mouse tumors in mice. 
Trypan blue is also nonenhancing, though it may produce a slight synergis- 
tic action given in conjunction with appropriate lyophilized tissues (17). 
Other data on specificity have been published by Casey and co-workers (4) . 

Studies employing differential centrifugation and filtration reveal an 
association of the enhancing agent or agents with the particulate com- 
ponents of the cell (5, 18). 

We now turn to another phase of the Jackson Laboratory program — the 
study of histocompatibility genes. These are defined as the genes which 
determine susceptibility and resistance to tumor and normal tissue trans- 
plants (3) . Special methods have been devised for this study, which may 
appropriately be called "mitogenetic," since they involve tissue grafts 
(transplantable tumors are employed in practice for various technical 
reasons), and such genetic techniques as the use of marker genes and the 
production of isogenic-resistant lines. These methods are quite distinct 
from classical serologic procedures. However, Hoecker, Counce and Smith 
(19), using these procedures in our laboratory, have fully confirmed our 
results for the histocompatibility-2 locus, which is also a blood-group- 
determining locus and hence amenable to studies of this type. 

For the sake of brevity, we shall give no description here of methods, 
which have been fully discussed elsewhere (3, 20, 21), but rather confine 
ourselves to a summary of certain results. 

We are particularly concerned here with the histocompatibility-2 locus. 
This locus can assume at least seven alternative or allelic forms, as follows 
(some of the strains known to carry each allele given in parentheses) : 
H-2 a 4 (strain A), H-2 d (strains BALB/c, C57BL/6Ks, B/10D/2, 5 DBA/2), 
H-2 k (strains C57BR/a, CBA, ST), H-2* (strains C57BL/6, C57BL/10, 
LP, 129/Rr), H-2 p (strain P), H-2 q (strain DBA/1), and H-2 r (strain 

These seven alleles may be divided into four groups on the basis of the 
presence or absence of one or both of two components (or histocompati- 

4 Allele H-2* was formerly called H-2 dk . The new designation has been agreed on by several investigators inter- 
ested in histocompatibility genes and blood group genes in the mouse. 

6 B/10.D/2 is an abbreviation for C57BL/10.DBA/2. This is a strain isogenic with C57BL/10 except for the fact 
that gene H-2 d , introduced from strain DBA/2 by an appropriate series of crosses, has been substituted for gene 

Vol. 15, No. 3, December 1954 


bility factors or antigenic factors), D and K. H-2 a carries both D and K 
H-2 d the D factor only, H-2 k the K factor only, and H-2\ H-2 P , H-#« and 
H-2 T neither D nor K. These factors were originally demonstrated by his- 
togenetic methods (20) , but they are equally demonstrable by appropriate 
antisera as red blood cell agglutinogens (19). They are therefore clearly 
antigenic in nature; in fact as compared with other isoantigens in the 
mouse they seem to give particularly high titers. 

Experimental: Methods and Results 

The demonstration by Kaliss that the enhancing effect can be produced 
by injection of normal tissue preparations, taken together with our analysis 
of the histocompatibility-2 locus, has made it possible to set up a precise 
analysis of the specificity of the enhancing phenomenon. In each test 
three strains are involved: 1) the host strain; 2) the strain providing the 
normal tissue used for the advance injections; 3) the strain providing the 
transplantable tumor. These strains were selected on the basis of the 
presence or absence of the D and K factors. The host strain was either D 
or K. The strains providing the normal tissue were from all four allelic 
groups, DK, D without K, K without D, and neither D nor K. The tumor 
was always DK. Host mice were given an appropriate series of lyophilized 
tissue injections, and subsequently inoculated with the living tumor. 
Uninjected controls were run with each experiment. 

The results are presented in table 1 . The first point of interest is that 
strain combinations involving the same H-2 allelic groups produced essen- 
tially similar results, particularly in any one experiment, even though the 
strains involved were genetically quite unrelated in other respects. Thus 
injection of D mice from two strains with tissue preparations from four 
strains, also D, (BALB/c, C57BL/6Ks, B/10.D/2, and DBA/2) produced 
no effect in any experiment, except for the death of 1 mouse out of 20 re- 
ceiving BALB/c kidney. Likewise, K mice receiving D tissue from two 
strains (BALB/c and DBA/2) in three experiments gave deaths in the pro- 
portions 12 of 20, 9 of 20, 10 of 20, and 16 of 19, a relatively homogenous 
distribution, with such fluctuations as there are apparently correlating 
with the relative enhancing potencies of liver, kidney, and spleen. 

As expected, the uninjected controls were negative or almost negative 
in each case. 

The results are summarized by allelic groups in table 2. It will be seen 
from this table that certain donor-host combinations produce clear 
enhancement (55 to 62 percent of the mice dying), while others either do 
not produce it (0 to 1 percent dying) or produce it to a weak degree only 
(21 to 22 percent dying). These distinctions are statistically valid. 

The significance of these results for the interpretation of the enhancing 
effect is indicated by table 3. It will be seen from this table that enhance- 
ment occurs when lyophilized tissue donor and tumor transplant donor 
share in common a histocompatibility-2 factor which is lacking in the host. 
Thus where the tissue shares K with the tumor, and the host is not K but 

Journal of the National Cancer Institute 




© <a 
S 8 

<^ t 

e - 



-2 e. 

2? *? 


qg »s£ 








o s 

t-h <m T-t a 

> S^r^^l 

7- •> ** 

oo o c 


/ — * — > 





T— 1 













































* — 





<N t-H 






i— 1 tH 






















































i— i 

I— 1 T-H t— 1 













- ^3 _t3 _ii CD 03 

1J TJ Tj <—* i-h 

^ ^ co 


„H -pH .|H O, q 


ON g 





www 3 









« * 


CO O* 











ffl o 

03 PQ 

s s 

"S * 

eg o 

.9 T3 


ss ^ 


s * 

§2, ^ 


•° P 

■i n 

2 M 

^ iS +a 

« ? fl 

lis I 

03 " J it-! S 

^ ft ft 

s « a 


o g c3 

I "a o 

ft^ a 


Ah « 

3 M 5 
s <n .y 
?™ -b 

^M I 
© .. d 

^ I § 


S co 


m <S V 

@2 i 

3 « 


^ a 

eg 03 

03 T3 

_. ,d a ^ 

Si s s 

§* ^ 2>- 

^ m g ft g 

m I a i i 

8 s 

10 d • 

.a -S -< 

03 © PQ 

I o^O 
~,d !C 


5 a j 

w © »0 

03 a o 

'd "a -9 

k^ ffl o3 
f*» ft (H 

•3 ^.a 

ftM >5 

•M -W > 

g .a 
& a ©■ 

y © o 

i : = 


B & * 

• g g 

2 P 

ra ^ © 

■81 I 

M 03 „ 

as 1 

•^ M d 



2 « S 

« d-s 
>> i "S 
-° 1 .1 

£ 2? .a 

S -d c 

a 11 ■§ 

« ftS 

as 2 

jg m © 


d o 


T3 X I 
© © tl 

!l S 

Is Q fe 

'O G- >» 

■2 S b 
aS 3 

1 M * 

& Q w 
H & 03' 

a -^ - 
I © § 



t» » .3 

I ^ -9 

•rt m ft 
o d 9 

© 03 O 

m a a <n 

8 1- ' "o3 -Sf 
t) © 

O © P 

8S >> 

" T3 bC ri 

® fl s 

a « a » 




Vol. 15, No. 3, December 19S4 



Table 2. — Percent mice dying from tumor {total number of mice 
in parentheses) grouped according to the presence or absence of 
histocompatibility factors D and K in host- and lyophilized tissue 
donor. Tumor in every case was DK 

Host geno- 

Donor genotype 


























D, enhancement occurs. Pretreatment with a foreign H-2 factor specifi- 
cally abrogates resistance to this factor. This result is exactly what would 
be expected in an immunizing procedure, with the important and curious 
qualification that the end result is not increased resistance to, but increased 
tolerance of, the test graft. 

Table 3. — Condensed summary of results* 












♦Tumor is DK; + most of mice died; — most of mice survived. 


Any discussion of the enhancing effect in the light of our present know- 
ledge must leave many questions unanswered, and must in some measure 
be tentative. Nevertheless, available data do suggest certain conclusions, 
and this would appear to be an appropriate time to present them. We 
shall venture three hypotheses, giving them in order of what seems to us 
descending assurance as to their validity. 

Hypothesis 1. — The {an) enhancing substance is a product of the H-2 locus. 
The evidence for this statement is the almost complete concomitance, in 
the host, tissue donor, and tumor donor system, as revealed in tables 2 
and 3, between H-2 genotype and the occurrence or nonoccurrence of 
enhancement. If a particular H-2 allele has to be present for enhance- 
ment to occur, we can see no escape from the conclusion that that allele 
or some product thereof is the cause of the enhancement. That it is a 
product of the H-2 gene and not the gene itself is proved by the fact that 
two cytoplasmic constituents, washed mitochondria and microsomes from 
appropriate strains and tissues, are potent enhancing agents (5). 

ial of the National Cancer Institute 


One point in this connection requires clarification. In the specificity 
experiments which we have described, a total of 10 different inbred strains 
were employed. A few of these have known common ancestry {e.g., 
C57BL/10 and C57BR/a) or are differentiated by a single gene difference 
{e.g., C57BL/10 and B/10.D/2), but most of them are quite unrelated and 
undoubtedly represent a wide diversity of genotypes. Besides the known 
differences at the H-2 locus, we can be sure that these strains are char- 
acterized by differences at a variety of other histocompatibility loci, 
even though the details of these differences are as yet mostly unanalyzed 
{21). How did it happen that virtually the only effect that emerged was 
the effect of the H-2 locus? Since the distribution of other histocom- 
patibility loci was in no way controlled, why did not the action of these 
loci completely confuse the picture? 

There is more than one possible answer to these questions. It may be 
that there is some peculiarity of the products of the H-2 alleles so that they 
can abrogate host resistance under appropriate experimental conditions, 
whereas other histocompatibility gene products have no such effect. Or it 
may be that the difference is quantitative rather than qualitative, and 
that the products of other loci also enhance, but to a lesser degree. 

There are two facts favoring this second alternative. We have pointed 
out elsewhere that the H-2 locus is a "strong" locus. We shall not take 
the time here to define "strong," other than to point out that it is simple 
and perhaps correct in the light of the work of Gorer [reviewed in (7)], and 
Hoecker, Counce and Smith {19), to think of a strong locus as one whose 
products are potent producers of isoantibodies. Readers interested in 
further details are referred to our earlier publications {20, 21). Suffice it 
to say here that the clear manifestation of the enhancing effect of the H-2 
substance, in a situation where it must have been competing with other 
histocompatibility gene products, may simply be a consequence of its 
relatively great potency. 

The second point to be noted is that there are certain irregularities in 
our data which can be interpreted as an enhancing action by other histo- 
compatibility gene products. An examination of table 2 will show that 
tissue from K and not-D-not-K strains tested in K hosts produced some 
enhancement as compared with the untreated controls, though the D 
factor expected to cause enhancement in this host genotype was lacking. 
The obvious interpretation is that other though weaker enhancing sub- 
stances, presumably also histocompatibility gene products, were present. 

It is perhaps worth mentioning here that establishment of isogenic 
resistant lines {3, 21) makes possible the design and execution of experi- 
ments in which enhancement, if it occurs, can be due only to the action of 
single histocompatibility genes or their products. We have in fact pro- 
duced 100 percent enhancement with strain A whole blood of a strain A 
tumor in A.SW mice, where the only genetic difference is at the H-2 locus 
(unpublished data) . 

A significant byproduct of our first hypothesis is the inference that in 
our chemical program for the isolation of the enhancing substance (s) 

Vol. 15, No. 3, December 1954 

672 proceedings: symposium on 25 years of 

we should be able to substitute for our present time-consuming assay 
based on the enhancing test a short serological assay for the H-2 antigen. 
Using either procedure, we should be assaying for one and the same 
compound. I am indebted to Dr. Mitchison for the suggestion that 
probably the easiest assay should consist of the absorption of an anti- 
serum against the donor H-2 allele with the fraction to be assayed, fol- 
lowed by an agglutination test of absorbed antiserum against donor red 

Hypothesis 2. — The enhancing effect is basically immunological. Five 
facts, already cited in some detail, which point to this conclusion are: 
1) the injections have to be made before the period of active transplant 
growth; 2) the effect persists; 3) the effect can be passively transferred; 
4) the effect is highly specific; 5) the enhancing substance, if our first 
hypothesis is correct, is a known isoantigen (Gorer, see review in 7). 
The mystery is why the effect is manifest, not as protection against the 
graft, but as increased tolerance thereof. 

As evidence that the contrast between the enhancing effect and normal 
immune phenomena is a real and significant one, a variety of observations 
may be cited. 

Increased resistance to, rather than increased tolerance of, tumor and 
normal tissue grafts can be evoked by appropriate types of pretreatment of 
the host (see review in 7) . This is possible with strain A tumor SAl and 
C57BL/6Ks and C57BR/a hosts, the combinations of tumor and host 
which we have used most frequently in the study of the enhancing effect. 
In transplantation studies the two types of end result therefore stand out 
in sharp contrast. 

Whereas the enhancing effect is passively transferred with serum, 
graft immunity is passively transferred only by cells and tissues, with 
lymph nodes the organs most clearly implicated (23, 24; see also review in 

While the enhancing effect is an unusual phenomenon, it is not without 
several close parallels in other areas of research. As we have pointed out 
elsewhere (10) it resembles in several respects Fel ton's (8) immunological 
paralysis and Anderson, Billingham, Lamkin and Medawar's (25) skin 
graft susceptibility in twin cattle. We shall not repeat this evidence here, 
but we do wish to point out that the argument has been strengthened by 
recent findings. The demonstration here presented of high specificity in 
the enhancing effect further emphasizes this feature of all three phenomena. 
Moreover, the recent brilliant demonstration by Billingham, Brent and 
Medawar (9) that mice can be rendered susceptible to skin homografts by 
injection as embryos with donor-strain tissues serves to close the experi- 
mental gap between the enhancing effect and the situation in cattle twins. 

All these facts point to the conclusion that under certain, as yet poorly 
defined, conditions the antibody-producing system responds, not by a 
reaction of immunity or resistance, but by an equally clear and specific 
genesis of tolerance. 

Journal of the National Cancer Institute 


If all these phenomena are fundamentally the same and fundamentally 
immunological, they deserve a common descriptive name. Billingham, 
Brent and Medawar (9) have called their induced graft susceptibility in 
mice "actively acquired tolerance." The name seems highly appropriate, 
and we propose that its use be extended to the enhancing effect. 

Hypothesis 8. — The enhancing response may be in part determined by 
the fact that the H-2 substance in our injected lyophilized tissue, while an 
effective isoantigen, is also a native mouse substance and is physiologically 
reacted to as such. Presumably any of the seven or more variant forms of 
the substance, corresponding to the seven or more H-2 alleles, can enter 
into, and adequately function in, some predetermined place in the cell's 
structure. A foreign variant of the substance, introduced in a host by 
experimental procedures, will compete with the native variant, just as 
antimetabolites of a vitamin compete with the structurally related normal 
vitamin in certain physiological processes (26). In our case, however, 
the competing substance is functionally adequate rather than inadequate 
and detrimental, and its incorporation in the cell, with resulting formation 
of a mosaic structure, may be predicted. An H-2 d mouse injected with 
H-2 k substance thus becomes partly an H-2 k mouse. 

A mosaic structure, though originally at least entirely at the inter- 
cellular rather than the intracellular level, is present in identical twin 
cattle and in Billingham, Brent and Medawar's graft-tolerant mice. 

Possibly also significant is the fact that both the pneumococcus poly- 
saccharides used to produce Fel ton's immunological paralysis, and the 
foreign strains of cells in twin cattle and graft-tolerant mice, persist, 
certainly for long periods, and probably throughout the life of the host. 
Persistence of antigen is said to be rare (27, p. 68). TVe have no evidence 
for the persistence of the enhancing substance following injection, but the 
postulate that it is incorporated into the host's cells would not imply 
rapid elimination. This question should be amenable to experimental 
answer in due course. 


When mice receiving homografts of certain transplantable tumors are 
given prior injections of lyophilized tumor or of lyophilized mouse kidney, 
liver, or spleen from the donor strain, progressive growth .of the tumor 
tends to occur. This phenomenon has been called the enhancing effect. 

A study of the specificity of the effect has been carried out. Hosts, 
normal tissue donors, and transplantable tumor were selected from strains 
with known genotypes with respect to the D and K factors of the histo- 
compatibility^ (H-2) locus. Enhancing experiments were carried out 
with these strains in various combinations. It was found that enhance- 
ment occurred if, and only if, tumor and normal tissue shared a factor 
which was lacking in the host (table 3) . 

The conclusion is drawn that the (an) enhancing substance must be a 
product of the H-2 locus. Keasons are presented for believing that the 
enhancing effect is immunological in nature despite its reversal of the usual 

Vol. 15, No. 3, December 1954 

316263—54 28 

674 proceedings: symposium on 25 years of 

immune effect, and the name "actively acquired tolerance" (after Meda- 
war and co-workers) is proposed. It is suggested that injected H-2 
substance becomes incorporated in the host's cells along with the native 


(1) Little, C. C: A possible Mendelian explanation for a type of inheritance 

apparently non- Mendelian in nature. Science 40: 904-906, 1914. 

(2) : The genetics of tumor transplantation. In the Biology of the Labora- 
tory Mouse, (Snell, G. D., ed.) Philadelphia, The Blakiston Co., 1941. 

(8) Snell, G. D.: Methods for the study of histocompatibility genes. J. Genetics 
49: 87-108, 1948. 

(4) Casey, A. E., Ross, G. L. and Langston, R. R.: Selective XYZ factor in C57 

black mammary carcinoma E0771. Proc. Soc. Exper. Biol. & Med. 72: 
83-89, 1949. 

(5) Snell, G. D. : Enhancement and inhibition of the growth of tumor homoiotrans- 

plants by pretreatment of the hosts with various preparations of normal and 
tumor tissue. J. Nat. Cancer Inst. 13: 719-729, 1952. 

(6) Kaliss, N.: Induced alteration of the normal host-graft relationship in homo- 

transplants of mouse tumors. Ann. New York Acad. Sc. In press, 1954. 

(7) Snell, G. D.: Transplantable tumors. In The Physiopathology of Cancer. 

(Homburger and Fishman, eds.) New York, Hoeber-Harper Book Co., 1953. 

(8) Felton, L. D.: The significance of antigen in animal tissues. J. Immunol. 61: 

107-117, 1949. 

(9) Billingham, R. E., Brent, L., and Medawar, P. B.: "Actively acquired 

tolerance" of foreign cells. Nature 172: 603-606, 1953. 

(10) Snell, G. D.: The immunogenetics of tumor transplantation. Cancer Res. 12: 

543-546, 1952. 

(11) Snell, G. D., Cloudman, A. M., Failor, E., and Douglass, P.: Inhibition and 

stimulation of tumor homoiotransplants by prior injections of lyophilized 
tumor tissue. J. Nat. Cancer Inst. 6: 303-316, 1946. 

(12) Kaliss, N., and Day, E. D.: Relation between time of conditioning of host and 

survival of tumor homografts in mice. Proc. Soc. Exper. Biol. & Med. 86: 
115-117, 1954. 
(18) Kaliss, N., and Molomut, N.: The effect of prior injections of tissue antiserums 
on the survival of cancer homoiografts in mice. Cancer Res. 12: 110-112, 

(14) Kaliss, N., Molomut, N., Harriss, J. L., and Gault, S. D.: Effect of pre- 

viously injected immune serum and tissue on the survival of tumor grafts in 
mice. J. Nat. Cancer Inst. 13: 847-850, 1953. 

(15) Kaliss, N., and Snell, G. D.: The effects of injections of lyophilized normal 

and neoplastic mouse tissues on the growth of tumor homoiotransplants in 
mice. Cancer Res. 11: 122-126, 1951. 

(16) Kaliss, N.: Effect of prior injection of non-mouse tissues on growth of tumor 

homoiografts in mice. Science 116: 279-280, 1952. 

(17) Kaliss, N., and Borges, P. R. F.: Effect of injected lyophilized tumor and 

trypan blue on host resistance to tumor grafts. J. Nat. Cancer Inst. 13: 
343-347, 1952. 

(18) Day, E. D., and Kaliss, N.: Investigations preliminary to physical and chemical 

characterization of substances in mouse tissues responsible for the abrogation 
of resistance in mice to tumor homotransplants. J. Nat. Cancer Inst. 15: 
000-000, 1954. 
(18) Day, E. D., Kaliss, N., Aronson, I., Bryant, B. F., Friendly, D.. Gabrielson, 
F. C, and Smith, P. M.: Investigations of substances in mouse tissues inducing 
alteration of normal host-homograft relationships. J. Nat. Cancer Inst. 15: 
145-159, 1954. 

Journal of the National Cancer Institute 


{20) Snell, G. D., Smith, P., and Gabrielson, F.: Analysis of the histocompatibility-2 
locus in the mouse. J. Nat. Cancer Inst. 14: 457-480, 1953. 

{21) Snell, G. D.: The genetics of transplantation. J. Nat. Cancer Inst. 14: 691-700, 

{22) Snell, G. D., and Borges, P. R. F.: Determination of the histocompatibility 
locus involved in the resistance of mice of strains C57BL/10-:c, C57BL/6-Z and 
C57BL/6Ks to C57BL tumors. J. Nat. Cancer Inst. 14: 481-484, 1953. 

{23) Brncic, D., Hoecker, G., and Gasic, G.: Immunity in mice against leukemic 
cells of the same genetic constitution. Acta U. Int. contre le Cancer. 7: 
761-764, 1952. 

{24) Mitchison, N. A.: Passive transfer of transplantation immunity. Proc. Roy. 
Soc, s.B, 142: 72-87, 1953. 

{25) Anderson, D., Billingham, R. E., Lamkin, G. H., and Medawar, P. B.: The 
use of skin grafting to distinguish between monozygotic and dizygotic twins 
in cattle. Heredity 5: 379-397, 1951. 

{26) Woolley, D. W.: Antimetabolites. Ann. New York Acad. Sc. 52: (Art. 8) 
1197-1378, 1950. 

{27) Burnet, F. M., and Fenner, F.: The Production of Antibodies, 2nd ed. Mel- 
bourne, Macmillan and Co., Ltd., 1949. 

Vol. 15, No. 3, December 19S4 



Dr. N. A. Mitchison, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

When Dr. Snell and his colleagues first described the effect of prior injection of 
lyophilized tissue on the growth of tumor homografts in 1946, one of the first question 
which came to mind concerned the specificity of the phenomenon. Was this effec 
another stimulation or inhibition of tumor tissue in general, or was he interfering with 
the delicate and highly specific immunological relationship between the host and the 
graft? Kaliss answered this question in one form, but we now have evidence of 
remarkable elegance from material of known genetical constitution. 

An answer in this form has been possible only because Dr. Snell has succeeded in 
identifying histocompatibility genes in so many of the standard inbred lines; and this 
in turn has been possible only because of the development by Dr. Snell of his isogenic 
resistant lines. One can hardly, I believe, appreciate the labor and the skill, immuno- 
logical as well as genetical, which has gone into the development of these lines, until 
one has seen Dr. Snell's laboratory in action. A large-scale program, which at its 
commencement offered no sure promise of success, has been systematically developed, 
and has produced magnificent results. The identification of many histocompatibility 
genes, and their isolation in isogenic lines, opens up possibilities in many directions. 
The demonstration of the specificity of the enhancing effect that we have just heard is, 
I believe, the first example of their use. The evidence that has been presented is surely 
convincing, but I think that Dr. Snell would agree that the final proof of his hypothesis 
will come through the use of the isogenic lines. In immunization or enhancing experi- 
ments, a difference between tissues from two lines, isogenic except at the H-2 locus, 
would indicate not only that the difference was due to the iso-antigens controlled by 
the H-2 locus, but also that no other iso-antigens were involved. 

It is a surprising observation that, among the various more or less unrelated inbred 
lines that have been investigated so far, the H-2 locus alone appears to control the 
specificity of the enhancing effect. It would be rash, however, to conclude that no 
other iso-antigenic differences exist or are of importance. If hemolytic disease of the 
newborn was the only method of detecting iso-antigenic differences in man, we should 
know about the Rhesus and Kell groups but not about the ABO group! 

Dr. Snell has made it clear that we do not have as much evidence about the mechan- 
ism of the enhancing process as about its specificity. However, what originally seemed 
to be a rather isolated phenomenon is now fitting into a wider biological context. Owen 
published his observations on the blood groups of dizygotic cattle twins a few months 
before the first observations on the enhancing effect were announced. Although at 
the time no connection was apparent, both Kaliss and Medawar have pointed out that 
what Medawar and his co-workers have termed "actively acquired tolerance'' probably 
involves a mechanism very similar to the enhancing effect. Medawar argues that 
because replacement therapy with normal or immune lymphoid tissue can restore an 
actively tolerant mouse to its normal condition, actively acquired tolerance must 
involve a specific immunological paralysis of the antibody producing system. We now 
have indications that similar replacement therapy is effective in the enhanced mouse. 
Recently, Owen has produced evidence that the surprisingly small fraction of Rh- 
negative women who become immunized against antigen from their offspring may be 
due to actively acquired tolerance: the mothers, when they were themselves in the 
uterus, were exposed to Rh antigen. All this will surely provide fresh stimulus for in- 
vestigation of the mechanism of enhancing. 


Journal of the National Cancer Institnte, Vol. 15, No. 3, December 1954 

The Expanding Knowledge of the Ge- 
nome of the Mouse * 

Margaret M. Dickie, Roscoe B. Jackson Memo- 
rial Laboratory, Bar Harbor, Maine 

Mouse genetics, which began shortly after the turn of the century follow- 
ing the rediscovery of Mendel's laws, has come into a golden age of devel- 
opment. The contributions of this small mammal have been most 
impressive in the study of development and physiological genetics which 
bear close relationships with medical research in all its aspects. 

There were, until the first of June 1954, 215 named genes and alleles 
in the mouse. Now according to latest reports there are 226 named 
genes and alleles. Trust laboratories have been designated on both sides 
of the Atlantic Ocean which maintain all useful mutations that otherwise 
might be discarded or lost. This laboratory now maintains about 100 
of these mutations. 

Hans Griineberg's The Genetics of the Mouse, a comprehensive atlas of 
the genetic and morphologic descriptions of all mutations, and the Mouse 
News Letter, a semiannual communication that is essential to keep abreast 
of current affairs in mouse genetics, have been used as the sources for this 
presentation of the chronology of the mouse genome. 

Before scanning the development of the genome it is well to review 
briefly the kinds of mutations that have been recognized over the years. 
One could go into the pleiotropic effects or the developmental action of 
genes that are under intensive investigation at the present time, but such 
a discussion is not warranted here. A loose classification of the types of 
mutations will suffice. It is realized that some mutations overlap these 
different categories but no final classification is available now. In respect 
to color, such as black or brown or dilution of color, there are 29 genes. 
In regard to hair texture such as curliness like rex or wellhaarigkeit, or 
absence of hair — like hairless and naked — there are 22 genes. Spotting 
patterns such as piebald, splotch and belted account for 22 genes. Skeletal 
defects number some 58 genes, among which are the anury series, undulated 
and Polydactyly. Seventeen genes account for eye and brain abnormalities 
such as microphthalmia and hydrocephalus. The behavior mutants, 
although properly listed as disorders of the nervous system, may be 
divided into two classes. One of these classes is the shaker syndrome 
or the choreic mutants; there are 18 of them such as waltzer, shaker and 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 28, 1954. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

680 proceedings: SYMPOSIUM ON 25 YEARS of 

jerker. In the other class are 9 epileptiform syndrome genes; mutations 
that cause convulsive rather than any circling' behavior, such as trembler, 
wabbler-lethal and agitans. Four genes cause anemias. Ten genes act on 
other physiological phenomena, e.g. dwarfing, the histocompatibility genes 
and obesity. About 35 genes are lethal in the homozygous condition (not 
many, compared with the 557 known in Drosophila by 1945). 

There are 10 recognized allelic series: albino series with 5 and possibly 
6 alleles; pink-eye series with 3 alleles; dilution series with 3 alleles; 
hairless series with 4 alleles; dominant spotting series with 3 alleles; the 
agouti series with 5 alleles; the brown series with 3 alleles; anury series 
with 23 alleles; histocompatibility-2 series with 10 alleles; and micro- 
phthalmia series with 3 alleles. 

Remutations have been reported in the literature most notably at the 
agouti and brown loci. A few remutations to rex and caracul have 
occurred. Five somatic mutations have definitely been recognized since 
they involved gonadal mosaics, but it is assumed that more have remained 

These are merely glimpses at the present state of affairs in mouse 
genetics and now a glimpse at the history of its development. 

The records show that albinism has been recorded and described since 
ancient times and the Japanese waltzing as a characteristic was known in 
the last century on other continents. However, piebald, symbol s, was 
described by Allen in 1903 shortly after the rediscovery of Mendel's laws. 
Following that, Bateson described dilution and in 1905 Cuenot's classic 
studies and conclusions on the lethality and inheritance pattern of the 
yellow gene was reported. There appears to be a silence until 1912 when 
Hagedoorn reported a silvering gene, but it is known that at that time 
C. C. Little, working as a student of Dr. W. E. Castle at Harvard, was 
fascinated with the potentials of different kinds of mice and was laying 
the foundation for the aristocrat of the inbred strains, the DBA. Follow- 
ing the announcement of the silvering gene, Rabaud reported the occur- 
rence of a luxate-like character. In England in 1915, the first vertebrate 
linkage was announced by Haldane, Sprunt and Haldane. The linkage 
was between albinism and pink eye. This linkage was independently 
reported in the United States by Castle and his co-workers. In the 
succeeding years this linkage was confirmed by other workers and as 
other alleles of the albino series appeared they were tested, extreme 
dilution by Detlefsen in 1921 and chinchilla by Feldman in 1922. Mean- 
while, in 1916 Little described the brown gene and the light-bellied agouti. 
Short ear was reported by Lynch in 1921. In 1923, Little and Bagg 
reported the occurrence of myelencephalic blebs, a gene still independent 
of all known linkage groups. Other genes reported shortly after were 
rodless retina, dominant spotting, otocephaly and waltzing. 

During this first period that has been arbitrarily set, the summary 
shows that about 15 investigators were concerned with mouse genes, one 
linkage group had been set up and 19 genes and alleles had been recorded 
in the literature (text-fig. 1A). 

Journal of the National Cancer Institute 

















Cc ch c e c'c° 



Ppp r 



A y A w Aa'o 





C ce h c« 



















Hrhrhr 1 " 

WW v w 





3 ANU 

| <t°" 


Text-figure 1A. — The chromosome map of 1903-1925. One linkage group was 
established in 1915. 

IB. — The physical map of the period 1926-1935. Genes marked with white 
squares were put on the map during this decade. Dates under linkage group 
number signify date first linkage of the group was reported. 

1C— The decade 1936-1945. 

Vol. 15, No. 3, December 1954 



O _0 

ji « n 


9 DCS E 

■ ■ 00 

5 2 E 

,o t 




• i CT> 


Q. CO O E > 

J] ■ 1 ■ ll^LJLJ Ii 


z o 




Ngn n 


n i i ■ n 



.c .e -c 

Ql Ql $ 

— <u 


, o «> — O 


X2 <J O O O 

— P-Q. 




B io a 

o o-ir 

1 1 

■ ft in 

Journal of the National Cancer Institute 



!§U§ 25 

g ftT3 O N_^» 

6s .d 

G, ft 



§T3 «S3 

els 8 




g ~0 

o d 

-t£ 03S 

§« SI 


£ 03 



3^3 il 

■IP'S a 

2-2 d t>-S 


g _ 


S 5 ? ^ 



«lC*l to 

5S 'L? h*. S. 5 

*- *» a rt 8 

") co cd 



3 dx, 

03 03 d- 


qs a fi'S w 

03 >, C3 



OS43^53 ,q3£ So.a.d'S 


d «"3 



O Cd 



s poo 

©<S © " 

iS > 6? 

g s 



^H o:d a? _ 

w d 

~" © 


o o 

o ^ ■ S rt Q. ffl 


§ s 

,-S^SS ^.2|^l^ 

.a "3 

||§ll|| |pll|| 

oft -S-d O-a ra-d 

Vol. 15, No. 3, December 1954 

684 proceedings: symposium 

In the next period 1926-1935, the number of workers in the field 
increased to about 30, and among them were Crew, Snell, Dobrovolskaia- 
Zavadskaia, Dunn, Murray and Griineberg. The second linkage group 
was reported by Dr. Gates in 1927 and by Dr. Snell in 1928. This linkage 
was between dilution and short ear. The number of genes named in- 
creased rapidly. By 1930 there were 31 and by 1935 there were 44, of 
which 14 had been placed on five linkage groups. These linkage groups 
were reported by Snell, Keeler and Roberts, and Quisenberry (text-fig. IB). 

The next arbitrary period (1936-1945) saw the number of linkage groups 
doubled, genes described totaled 81, and 55 were located on these 10 
linkage groups. Investigators pursuing this phase of research had risen 
to about 50 and included among others, Fisher, Her twig and MacDowell 
(text-fig. IB). 

Now at least 90 geneticists are actively engaged in genetics research 
with this rodent. Names added to our distinguished list over this decade 
must include at least Russell, Gluecksohn-Waelsch, Heston, Carter, 
Falconer, Green, Garber and Griffen. Since 1945, 145 new genes have 
been reported which brings the total to 226. One hundred and sixteen 
of these named genes and alleles are now located on 15 linkage groups. 
Three new linkage groups were established and 31 genes were located in 
the past year alone (text-fig. 2). 

The event which I am sure those working with mouse genes have been 
eagerly awaiting for many years, became a fact last year, i.e., the identifi- 
cation of several sex-linked genes and their established linkage relation- 
ships on the sex chromosome (XX). Mottled was described by Fraser 
in 1951, and brindled by Falconer in 1952. Garber reported on bent tail 
in 1952. The linkage of bent tail and tabby, and brindled and mottled, 
were recorded last year and early this year jimpy was located on the 
group. Two other genes, a sex-linked lethal and tortoise shell are sex- 
linked but no linkage data to establish their position on the map is avail- 

With increasing numbers of geneticists working in this field and the 
number of mutations being discovered and studied, it appears that very 
soon all 20 linkage groups will be identified in this "golden age" of mouse 

Session V. Carcinogenesis in Endocrine 
Organs (Roundtable Discussion) 

Chairman, Dr. Frank E. Adair, Attending Sur- 
geon Emeritus, Memorial Hospital, New York, 
N. Y.; President, Board of Trustees, Roscoe B. 
Jackson Memorial Laboratory 

Contributors : 

Dr. Jacob Furth: Thyroid-Pituitary Tumorigenesis 
Dr. W. U. Gardner: Studies on Ovarian and Pituitary Tumorigenesis 
Dr. Katharine P. Hummel: Induced Ovarian and Adrenal Tumors 
Dr. George W. Woolley: Carcinogenesis in the Adrenal 

Floor discussion leaders: 
Dr. W. B. Atkinson, Dr. E. Elizabeth Jones*, Dr. Flavia 
Richardson, Dr. Robert Speirs 

•Discussion not submitted. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Thyroid-Pituitary Tumorigenesis h 

Jacob Fukth, The Children's Cancer Research 
Foundation, Boston, Mass. 

This introduction to the roundtable discussion aims to summarize and 
evaluate recent observations on the mechanism of induction and character 
of thyrotrophic in comparison with other types of pituitary tumors. 
Most experimental data have been or will be fully published elsewhere; 
therefore only a summary, discussion and references of this presentation 
will be printed here. 

There is a reciprocal relationship between the function of the thyroid 
and the thyrotrophs of the pituitary and it is possible to produce tumors 
of either the thyroid or the pituitary by interfering with their physiologic 
equilibrium. When the synthesis of the thyroid hormone (TH) is blocked 
without interfering with growth responsiveness of the thyroid epithelium 
(as accomplished by goiterogens), thyroid tumors will result (1-8). 
Invasive thyroid tumors occurring naturally in fish (4) and rats (5) are 
probably due to dietary iodine deficiency. Thyroid tumors induced in 
mice by thiouracil have been shown to be transplantable to hosts whose 
thyroid function had been depressed by a goiterogen, but not to normal 
animals. Upon successive transplantation such dependent thyroid 
tumors give rise to autonomous neoplasms (6) . However, thyroid tumors 
can also be produced in normal hosts with sustained stimulation of the 
normal thyroid by excessive quantities of the thyroid-stimulating hormone 
(TSH) . This is easily accomplished by grafting a thyrotrophic pituitary 
tumor on a normal host (7). 

Destruction of the thyroid of mice, best accomplished by administration 
of radioiodine (8), induces pituitary tumors. Ionizing radiation was 
thought by Gorbman and Edelmann (9) to be essential for the induction 
of these tumors, but experiments of Dent et al. (10) have shown that 
surgical thyroidectomy will accomplish the same; other experiments in 
the same laboratory indicate that the tumors induced by radiothyroidec- 
tomy are composed of thyrotrophs and that radiation plays a minor role, 
if any, in the induction of thyrotrophic pituitary tumors (11). This idea 
is supported by recent experiments of Moore et al. (12) indicating that 
long continued administration of propylthiouracil will also induce pituitary 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 29, 1954. 

a The investigations of the author are being supported currently by the National Cancer Institute: sources of 
support of earlier work are indicated in the original reports. 

Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

688 proceedings: symposium on 25 years of 

tumors. The transplantability and character of the tumors induced by 
this compound has not been established but it is probable that they are 
induced by sustained deficiency of TH, as are the pituitary tumors which 
develop following thyroidectomy. It can therefore be assumed that these 
pituitary tumors are also composed of thyrotrophs. That the pituitary 
tumors induced by surgical- and radio-thyroidectomy are identical in 
character has been shown by Dent et al. (10). The precise conditions 
required for induction of thyroid tumors by propylthiouracil on the one 
hand, and pituitary tumors on the other, remain to be determined. It 
appears that the determining factor is the degree of block of synthesis 
of TSH — partial block yielding thyroid tumors; more complete block, 
pituitary and thyroid tumors; and complete destruction of the thyroid 
epithelium, pituitary tumors only. 

In the course of successive transplantations, thyrotrophs tumors in- 
duced by thyroidectomy go through two major phases: a) conditioned or 
dependent, and b) autonomous (18). All of more than 13 primary 
thyrotrophs tumors studied by serial transplantations in normal and 
athyroid hosts were dependent in the first transplant generation, but 
after the second or later passage, most of them turned autonomous. With 
one tumor strain the dependent character of the tumor has been main- 
tained through six consecutive passages made in the course of 2 years. 

Evidence for the occurrence of successive changes in both dependent 
and autonomous tumor cells is suggested by the increasing growth rates 
of the tumors, by the increasing histologic and cytologic dedhTerentiation 
of the tumor cells, and by diminishing hormone production (11, 18). The 
fact that the first generation grafts never took in normal hosts indicates 
that the autonomous variants arose from dependent cells by a transfor- 
mation analogous to a somatic mutation. It is probable that the ac- 
celerated growth rate occurring in the course of successive passages of 
both dependent and autonomous tumors is caused by similar modifica- 
tions of the tumor cells, but assays of single-cell progeny are required to 
prove this supposition. 

Similarly, it remains to be shown whether the conditioned neoplasms 
are composed of unaltered normal cells, growing progressively in response 
to a tremendous increase of their normal physiologic stimulant, or of 
somewhat altered but still responsive cells. The induction of pituitary 
tumors can be prevented by administration of thyroid hormone (14, 15) 
and the take of a transplanted dependent tumor can be prevented by a 
similar procedure. Furthermore, it has been possible to restrain, but not 
to arrest, already growing dependent pituitary tumors by administration 
of thyroid hormone. It remains to be determined whether this failure of 
complete regression is due to difficulties of administering large, ' 'neutral- 
izing" quantities of thyroid hormone or to a diminished responsiveness 
of the dependent tumor cells. 

In addition it would be desirable to determine whether or not the 
primary dependent tumor cells would regress in the original thyroidec- 
tomized hosts following administration of TH and, if not, the exact time 

Journal of the National Cancer Institute 


when responsiveness to TH is lost. Tumorigenesis by derangement of 
the normal hormonal equilibrium between thyroid and pituitary could 
serve as a model for research on neoplastic growths of other cells because 
of the relative ease and precision of quantitating the two opposing forces, 

Complete destruction of the thyroid gland is not essential for the induc- 
tion or growth of dependent pituitary tumors. Kesidual thyroidal epi- 
thelium, present either after surgical or radio thyroidectomy, appears to 
be unable to proliferate sufficiently to produce enough hormone to restrain 
a dependent tumor once the balance is tilted in favor of tumor growth (10). 
Failure of thyroid remnants to undergo adequately compensating hyper- 
plasia at the site of radiothyroidectomy can be explained by the sclerotic 
changes seen following irradiation, but it is puzzling why regeneration of 
the thyroid remnants after subtotal surgical thyroidectomy does not pro- 
ceed far enough to restrain the normal balance between thyroid epithelium 
and thyro trophic pituitary cells. The possible explanations include: 
1) Regeneration is limited by the surrounding tissues. The bulk of the 
regenerated thyroidal tissue in surgically thyroidectomized animals never 
reaches that of the normal thyroid. 2) An alteration occurs in the respon- 
siveness of the cells or of the quality of the hormone produced during a 
sustained uncompensated period. 

Further evidence favoring the view that radiation plays no role in the 
induction of thyrotrophs tumors in radiothyroidectomized mice is fur- 
nished by the observations that pituitary tumors induced by whole-body 
irradiation are different in character from those induced by radiothyroidec- 
tomy (16). Of 10 such tumors assayed by means of grafts on animals of 
the strain of origin only one had some thyrotrophic potencies and even 
this was accompanied by growth-promoting potencies. Three of the 
tumors assayed yielded pure adrenotrophic transplantable tumors and 
three others appeared to have only mammary-gland stimulating effect. 
All seven tumors induced in mice by ionizing irradiation were autonomous 
at the start (17). 

While all of at least five strains of mice thus far tested by different 
investigators readily developed pituitary tumors following radiothyroidec- 
tomy, the rat proved resistant to this type of tumorigenesis. It is not 
known whether this difference is a species characteristic, and if so, whether 
normal diet supplies sufficient TH to thyroidectomized rats to prevent the 
development of pituitary tumors. 

It is not known whether pituitary tumors will develop in man in the 
absence of the thyroid; older records are inadequate and recently compen- 
sation therapy by administration of TH counteracts such tendencies. On 
the other hand, it is certain that sustained excess of thyrotrophs, brought 
about by goiterogens, play a significant role in the development of both 
adenomas and carcinomas of the thyroid in man. 

The occurrence of pituitary tumors in rats and mice following adminis- 
tration of estrogen was reported in 1936 independently from three labora- 
tories (18). This has been amply confirmed, but thus far the character 

Vol. 15, No. 3, December 1954 
316263—54 29 

690 proceedings: SYMPOSIUM ON 25 YEARS of 

of the tumors so induced has not been ascertained. Dunning (19) has 
shown in the rat, and Gardner (20) in the. mouse, that such estrogen- 
induced pituitary tumors are transplantable to hosts similarly conditioned 
by administration of stilbestrol but the nature of the hormonal secretion 
of these tumors is a matter of mere speculation. It is highly probable 
that they are gonadotrophic but equally good arguments can be presented 
for their being either luteotrophic or follicle stimulating or lactogenic. 

Experimental material is now available to study mammotrophic tumors. 
The pituitary tumors developing in gonadectomized mice (21) were asso- 
ciated with extensive alveolar hyperplasia of the mammary gland but this 
was noted also in mice with primary thyrotrophic pituitary tumors and 
the evidence thus far indicates that the thyrotrophic tumors themselves do 
not secrete prolactin. Mammary-gland stimulation is not a characteristic 
feature of grafted thyrotrophic tumors (7) and assays by Bates failed to 
disclose lactogenic properties (11). 

Three transplantable autonomous pituitary tumor strains derived 
from mice that had been exposed to whole-body irradiation invariably 
stimulate the mammary glands without causing an enlargement of other 
endocrine organs (17). 

The discovery of a conditioned phase of tumor growth preceding the 
autonomous phase, deserves special attention because the alteration in 
the former resides in the host and is controllable by restoration of hor- 
monal deficiency even though the tumor progresses and metastasizes in 
conditioned hosts until death ensues. Since all cells have regulatory 
forces, it is highly probable that some cells, other than those of the endo- 
crine organs, have a conditioned phase preceding the autonomous one. 
Furthermore, it is of importance from both the theoretical and the prac- 
tical standpoint that following or at the time of acquisition of autonomy, 
the tumor cells may acquire some degree of dependency on the physiologic 
agent which originally restrained them. 

The pituitary tumorigenesis of thyrotrophs furnishes excellent material 
to study the various factors involved in the transformation of a normal 
into a highly autonomous cancer cell. 


(1) Purves, H. D., and Griesbach, W. E.: Studies on experimental goitre. VII. 

Thyroid carcinomata in rats treated with thiourea. Brit. J. Exper. Path. 27: 

294-297, 1946. 
{2) Bielschowsky, F., Griesbach, W. E., Hall, W. H., Kennedy, T. H., and 

Purves, H. D.: Studies on experimental goitre: The transplantability of 

experimental thyroid tumours of the rat. Brit. J. Cancer 3:541-546, 1949. 

(3) Morris, H. P., and Green, C. D. : The role of thiouracil in the induction, growth, 

and transplantability of mouse thyroid tumors. Science 114: 44-46, 1951. 

(4) Schlumberger, L. : Personal communication. 

(5) Bielschowsky, F.: Chronic iodine deficiency as cause of neoplasia in thyroid 

and pituitary of aged rats. Brit. J. Cancer 7: 203-213, 1953. 

(6) Morris, H. P., Dalton, A. J., and Green, C. D.: Malignant thyroid tumors 

occurring in the mouse after prolonged hormonal imbalance during the inges- 
tion of thiouracil. J. Clin. Endocrinol. 11: 1281-1295, 1951. 

Journal of the National Cancer Institute 


(7) Furth, J. : Morphologic changes associated with thyrotrophin-secreting pituitary 

tumors. Am. J. Path. 30: 421-463, 1954. 

(8) Gorbman, A.: Tumorous growths in the pituitary and trachea following radio- 

toxic dosages of I 131 . Proc. Soc. Exper. Biol. & Med. 71: 237-240, 1949. 

(9) Gorbman, A., and Edelmann, A.: Role of ionizing radiation in eliciting tumors 

of the pituitary gland in mice. Proc. Soc. Exper. Biol. & Med. 81: 348-350, 

(10) Dent, J. N., Gadsden, E. L., and Furth, J.: On the relation between thyroid 

depression and pituitary tumor induction in mice. Cancer Res. In press. 

(11) Furth, J., Dent, J. N., Burnett, W. T., Jr., and Gadsden, E. L.: The mecha- 

nism of induction and characteristics of pituitary tumors induced by thyroidec- 
tomy. J. Clin. Endocrinol. In press. 

(12) Moore, G. E., Brackney, E. L., and Bock, F. G.: Production of pituitary tumors 

in mice by chronic administration of a thiouracil derivative. Proc. Soc. 
Exper. Biol. & Med. 82: 643-645, 1953. 

(13) Furth, J.: Conditioned and autonomous neoplasms: A review. Cancer Res. 

13: 477-492, 1953. 

(14) Goldberg, R. C, and Chaikoff, I. L.: Development of thyroid neoplasms in 

the rat following a single injection of radioactive iodine. Proc. Soc. Exper. 
Biol. & Med. 76: 563-566, 1951. 

(15) Gorbman, A.: Factors influencing development of hypophyseal tumors in mice 

after treatment with radioactive iodine. Proc. Soc. Exper. Biol. & Med. 80: 
538-540, 1952. 

(16) Furth, J., Gadsden, E. L., and Upton, A. C: ACTH secreting transplantable 

pituitary tumors. Proc. Soc. Exper. Biol. & Med. 84: 253-254, 1953. 

(17) Furth, J., and Gadsden, E. L.: To be published. 

(18) Gardner, W. U., Pfeiffer, C. A., Trentin, J. J., and Wolstenholme, J. T.: 

Hormonal factors in experimental carcinogenesis. In The Physiopathology 
of Cancer (Homburger, F., and Fishman, W. H., eds.). New York, Hoeber, 
1953, p. 256. 

(19) Dunning, W. F., Curtis, M. R., and Segaloff, A.: Strain differences in response 

to diethylstilbesterol and the induction of mammary gland and bladder cancer 
in the rat. Cancer Res. 7: 511-521, 1947. 

(20) Gardner, W. U. : Studies on ovarian and pituitary tumorigenesis. J. Nat. Can- 

cer Inst. 15: 693-709, 1954. 

(21) Dickie, M. M., and Woolley, G. W.: Spontaneous basophilic pituitary tumors 

of the pituitary glands in gonadectomized mice. Cancer Res. 9: 372-384, 1949. 

Vol. 15, No. 3, December 1954 

Studies on Ovarian and Pituitary Tu- 
morigenesis lf 2 

W. U. Gardner, Department of Anatomy, Yale 
University School of Medicine, New Haven, Conn. 

Almost 21 years ago I was first introduced to an export of Maine, the 
inbred mouse. After he had moved to Connecticut, Dr. L. C. Strong 
introduced me to about 2,000 of these mice in August 1933. He extolled 
their virtues; the unique value of having many almost identical animals 
or, considering different strains, many animals uniformly unalike, with 
which to perform experiments. 

Mice of some of the strains had, at that time, known differences in 
susceptibility to mammary cancer so the comparative mammary develop- 
ment and structure were first studied. It was during the course of these 
studies that a most unusual mouse was found. One of the last two mice 
of Dr. Strong's EI strain had multiple mammary tumors, lactating mam- 
mary glands, a cystic and hyperplastic endometrium, bilateral granulosa- 
cell tumors, nodular, hyperplastic adrenal cortices, and a chromophobic 
adenoma of the pituitary gland (1). It was really a "tumor" mouse. It 
started a long series of investigations on tumors of the ovaries and the 
pituitary glands. It was to an experimental endocrinologist interested 
in tumors what the free martin was to those interested in the determina- 
tion of sexual differentiation. 

For a number of years investigations were undertaken in which attempts 
were made to reproduce tumors of these types — not without considerable 
success. Multiple tumors could be obtained in mice by suitable combina- 
tions of hormonal and genetic influences, but we were never able to obtain 
the entire combination of tumors noted in the one mouse mentioned above. 
Attempts to reproduce by more controlled experimentation that which 
can occur spontaneously has not been entirely successful. 

We assumed, at that time, that the tumors resulted from an imbalance 
of hormones and gradually devised an outline for study of qualitatively 
and quantitatively controlled imbalances (2). 1) Modification or differ- 
ences of rate of hormone production; 2) modifications or differences of 
rates of hormone inactivation, excretion or utilization; 3) modification or 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 29, 1954. 

3 These studies have been reviewed, with more adequate credit to other investigators who have contributed to 
this field, in the general references given at the end of the paper. The original investigative work reported here 
has been supported by grants from the Jane Coffin Childs^Fund for|Medical Research, the National Cancer 
Institute of the National Institutes of Health, U. S. Public Health Service, and the Anna Fuller Fund. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

694 proceedings: symposium on 25 years of 

differences of the sensitivity of end organs, or 4) differences of the quality 
of hormones produced may result in hormonal imbalances or differences 
of an endocrine nature among mice of diverse strains. These were not 
fictitious possibilities; they all had been observed, by interpretation at 
least, although not necessarily all in any one animal or species. In part, 
hormonal imbalances have been reproduced in two ways under somewhat 
carefully controlled conditions: 1) The administration of specific exogen- 
ous hormones, or hormone-like substances, can alter or augment rates or 
periodicities of end-organ response or of hormone production; and 2) ef- 
fective production or destruction of intrinsic hormones may be modified 
through operative procedures. 

The above four major variables can in turn be influenced in some in- 
stances by a) genetic differences, b) nutritional influences, c) ontogenic 
influences, d) production of antagonistic or augmenting substances, e) dis- 
ease, f) etc. The total number of variables is enormous; their analysis is 
difficult, slow, and tedious. The boundaries between some of the quali- 
ties mentioned above are difficult to define sharply. What is interpreted 
to be an augmented rate of hormone production in reality may be merely 
an augmented or unusually high sensitivity of the end organ used to esti- 
mate the level of hormone activity. For example, Dr. Trentin (3) found 
that the vaginal epithelium of strain C57 mice was about 5 times as 
sensitive to injected estrogens as was the vaginal epithelium of strain A 
mice; mice of other strains and hybrids revealed intermediate sensitivities. 
Also, the mammary glands of mice of some strains were more sensitive to 
stimulation by estrogens than were those of other strains (4). In these 
instances, the end organ itself seemed more sensitive in mice of some strains 
than others, but is this true? May not mice of some strains be inactiva- 
ting estrogens more rapidly or counteracting the effects of these estrogens? 
What at first seems to be a simple difference in sensitivity may not be. 

Strain C57BL mice in our laboratory acquire pituitary tumors when sub- 
jected to prolonged treatment with estrogenic hormones, at least estrogens 
such as estrone and estradiol (and its esters) and stilbestrol (5). Mice 
of only one other inbred strain in our laboratory frequently show such 
tumors when similarly treated. The pituitary glands of all mice undergo 
some hypertrophy when estrogens are injected. Adult female mice have 
larger pituitary glands than males, indicating a stimulating effect of in- 
trinsic hormones. The differences in size are slight, the hypertrophic 
glands of estrogen-treated mice of most strains rarely exceeding 4 mg. 
or approximately twice the size of the control glands. The tumors may 
attain very large size, are localized or nodular in origin, appear to arise in 
the hypertrophic glands. 

The tendency for mice of strain C57BL to acquire pituitary adenomas 
when exposed to extrinsic estrogens is transmitted to their hybrid des- 
cendents (6). From 75 to 83 percent of the estrogen-treated (CBA X 
C57BL)F! hybrids had, at death, pituitary glands exceeding 12 mg. 
Smaller glands also occasionally showed nodular enlargements but since 
not all of them were studied histologically, for various reasons, the term 

Journal of the National Cancer Institute 



tumor was applied only to the larger glands that almost uniformly con- 
tained extensive areas of, or were composed entirely of, abnormal, non- 
granular chromophobic cells of a type not found in the normal gland. 
Both male and female mice transmitted the tendency for pituitary tumors 
to appear in their estrogen-treated hybrids. The tumors appeared earlier 
in males than in females. Some attained very large size, almost 300 
mg., and distorted the shape of their hosts' calvaria. Few tumors 
occurred in mice treated for less than one year. Other hybrid mice 
(A X C57BL and C3H X C57BL) also acquired such tumors (tables 1, 

2, and 3). The incidence of pituitary tumors was much lower, only 21 
percent, when hybrids (CBA X C57BL) were backcrossed (text-fig. 1, 
table 2) to mice of the CBA strain, and was 63 percent when backcrossed 
to mice of the C57BL strain (7). The tumors also occurred approxi- 
mately 100 days earlier in mice of the latter group (table 2, text-figs. 2, 

3, and 4). If adequate correction for ages of survival had been made, 
mice of the latter group would have had a relatively much higher incidence 
than mice that carried less C57BL chromatin. The transmission of the 


1 — Pituitary tumors among hybrid and backcross mice 
estrogens for prolonged periods 

that received different 

Continuous treatment until death 

Treatment stopped before autopsy- 




ber of 

of treat- 


Size of pi- 


Age of 

mice at 



Period of 

Size of pitu- 






B 16.6 
B 25 


B 16.6 
B 25 

B 25 

B 16.6 








20. 3 





16. 5-120. 5 

12. 0- 42. 3 

20. 3 

20. 0- 79. 8 

9. 0-118. 5 
15. 0-117. 5 

21. 0-138. 

27. 3-281. 5 

28. 5- 44. 3 

22. 3- 47. 3 




16. 5-111. 3 

cc 2 



41- 49 

47. 5-215. 





32- 61 



51. 5- 53. 5 

cc 3 




17. 3- 53. 3 

A 72 




18. 3 

cc 5 



14. 0-132. 5 
12. 5- 44. 

*CCi(C57BL9 XCBAc?);CC 2 (CBA$ X C57BW); HCi (C57BL9 X C3Hd"); CC 3 ( [CBA 9 X C57BLc?] 
X CBAo"); A72 (C57BLCJ 1 X A 9); CC 5 ( [CBA 9 X C57BW1 X C57BLcf). 
t SS— 25 mg. stilbestrol weekly. 

DPB — 25 fig. estradiol dipropionate in sesame oil weekly. 

ES— pellets of estrone subcutaneously. 

B 16.6 — 16.6 Mg- estradiol benzoate in sesame oil weekly. 

B 25—25 Mg- estradiol benzoate in sesame oil weekly. 

Vol. 15, No. 3, December 1954 



tendency for pituitary tumors was compatible with a dominant genetic 
factor or factors as regulating influences. 

Table 2. — The incidence of pituitary tumors among male estrogen-treated hybrid and 

backcross mice 



Age with tu- 



tary tumors 


age at 


Origin of group 



death — 














C57BL9 X CBAd" 








cc 2 * 

CBA? X C57BLcf 








cc 3 

(CBA 9 X C57BLc?) 

9 X CBAcT 








cc 4 

(C57BL9 X CBAd") 

9 X CBAcf 








cc 5 

(CBA 9 X C57BLc?) 

9 X C57BLc?) 








* Data published in Cancer Res. 1: 345, 1941. 

Table 3. — The incidence of pituitary tumors among estrogen-treated hybrid mice derived 
from the C57BL and CSH strains 


Origin of group 



With pitui- 
tary tumors 

Age with tu- 


age at 

death — 






HC 2 

C57BL 9 X C3H cf 
C3H9 X C57BW 
(B)* C3H9 X 







67. 1 


514. 2 



HC 3 




C3H— without the mammary tumor agent. 

P C 57 £ X CBA(? CBA$XC 57< ? 


Text-figure 1. — Scheme of one group of reciprocal hybrid and backcrossed mice. 
The incidence of pituitary gland tumors among the estrogen-treated male mice of 
these groups is given in table 2. No pituitary gland tumors were noted in the un- 
treated controls. 

Journal of the National Cancer Institute 













, 3 -(CBA ? 
-16.6 or 

xc 57 4> X e 

itradiol b 


enzoate w 



38<£- • 











o • 
• •o 


> • 
••• o • 

•o • • 
»«e o • 

• • • 




350 400 450 

Age in days at death 




Text-figure 2. — Age and weight distribution of pituitary gland or tumors of back- 
crossed mice. Similar data for the hybrids have been published (6) . 

Hybrids derived from reciprocal crosses of mice strains C57BL and C3H, 
when given estrogens, had pituitary tumors when killed at average ages 
in excess of 500 days (table 3, text-figs. 5 and 6). The apparent maternal 
influence of strain C3H was indirect rather than direct; many of the mice 
with the mammary tumor agent died early with mammary tumors and 
hence did not live long enough to acquire pituitary tumors (table 3). 

Pituitary tumors, according to the definition given above, were observed 
in none of the untreated mice of these strains or hybrid groups. 

The pituitary tumors are transplantable; about one of five has grown 
when transplanted subcutaneously into estrogen-treated mice; they have 
not grown in untreated mice. Even those that did grow have not grown 
in all of the estrogen- treated mice into which they have been transplanted. 
Because a small number of hosts were used it is difficult to compare the 
relative transplantability of these tumors with those of other endocrine 

Vol. 15, No. 3, December 1954 







,1 60 

* 40 





< C 57? X 

CBAd*) ^ 


rodiol ber 

izoate weekly 

9 ? -o 








a o 


o _ o •• 
1 •• • 

• •• 

oo • 

• • 


• •• 

• • • 






300 350 400 

Age tn days at death 

Text-figure 3. — Age and weight distribution of pituitary glands or tumors of back- 
crossed mice. Similar data for the hybrids have been published (6). 

glands. The transplanted tumors attained detectable proportions only 
after prolonged periods, usually one year or more, and almost always at a 
time when the host had a pituitary tumor developing in its own calvarium. 
After their appearance they grew slowly and progressively to attain diam- 
eters in excess of 2 cm. No tumor could be carried for more than four 
transfer generations. The very delayed initiation of palpable growths in 
estrogen-treated animals made it impossible to carry large numbers of 

The tumors were not reversible in the animals of origin insofar as could 
be determined, that is, they continued to grow when estrogen- treatment 
was stopped. Although it was impossible to ascertain the sizes of the 
tumors, except by changes in the shape of the calvaria, after the cessation 
of hormone treatment, tumors removed for up to 140 days after the cessa- 
tion of estrogen treatment showed no evidence of regression and had 
apparently continued to grow (table 1; text-figs. 7 and 8). 

Journal of the .National Cancer Institute 





O 80 


O 50 


cc 5 


(CBA$ x 
43 mic 

c 57 <?) 5 

e-16.6 or 

x C g7 o 4 

25.0/jg es 


trodiol ber 


zoate we 


2j- o 
41 /-• 























uv c 

- «. 

• 4 


250 300 350 400 450 500 550 600 650 

Age in days at death 

Text-figure 4. — Age and weight distribution of pituitary gland or tumors of back- 
crossed mice. Similar data for the hybrids have been published (6). 

At this time it is well to emphasize the possibility that the environment 
promoting tumorigenesis may or may not be the same as that necessary 
for tumor growth. Once, however, the environment has been established 
to promote growth of a dependent tumor, and it seems that temporal 
factors are of importance, it is not reversible. It might be assumed that 
the foci of tumorigenesis occur spontaneously with advancing age in 
animals with the proper genetic constitution but fail to grow because the 
hormonal environment is inadequate. If this is true then tumors should 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 


45 m' 

1 1 


ce - 16.6 or 25 }jg. estradiol benzoate weekly 


l2o- o 
33 c?- • 







| 80 






5 60 










• c 


( g&* 

o • 






J o 


400 450 500 550 600 

Age in days at death 



Text-figure 5. — The age and the Weight distribution of pituitary-gland tumors in 
hybrid mice (C57BL $ X C3H <?). 

appear more quickly when old mice are given exogenous estrogens. This 
has not been studied adequately. However, if such an assumption were 
true it would be even more difficult to explain the prolonged quiescence of 
the transplants of tumors in estrogen- treated mice. The growth of trans- 
plants of tumors and of the original neoplastic foci tend to parallel one 
another indicating that the growth-promoting environment is slowly 
created; that it depends on prolonged exposure to estrogens. 

What is responsible for the origin of these tumors? Actually we do 
not know. We do know that estrogens do produce changes in the func- 
tion of the anterior pituitary by reducing its gonado trophic content and 
growth-hormone content. It is possible that estrogens prevent the 
synthesis of certain pituitary hormones and the pituitary gland attempts 
to overcome this deficiency by adding new cells. After prolonged periods 
some of the cells among this population may lose all regulation by the 
normal restricting influences and nodular overgrowths or adenomas result. 

Dr. Furth (8) has indicated that somewhat analagous circumstances 
may predispose to pituitary tumors in thyroid-deficient mice and has 

Journal of the National Cancer Institute 



HC 3 



mice — 

16.6 pg. estradiol benzoate weekly 





.1 <?-• 












'5 60 








• • 










Age in days at death 

Text-figure 6. — The age and weight distribution of pituitary-gland tumors in hybrid 
mice [C3H 9 (without the mammary tumor agent) X C57BL cF]. Hybrids born 
of C3H mice with the mammary tumor agent did not live long enough to acquire 
pituitary gland tumors — they died with mammary tumors when treated with 

described the "feed-back" inter-regulation of thyroid and pituitary func- 
tion. Essentially this is the same hypothesis as that stated above. 

Granulosa-cell tumors occasionally occur in untreated mice that are 
not as involved in a polyglandular syndrome as indicated above; their 
incidence is low in mice of most strains. They were first observed at a 
relatively high incidence by Dr. Furth (9) in mice subjected to roentgen 
irradiation. Roentgen irradiation destroys the ova and reduces or modi- 
fies the endocrine function of the ovary. Subsequently ingrowths occur 
from the germinal epithelium that can be associated with renewed evi- 
dence of some endocrine activity (10). Granulosa-cell tumors first ap- 
peared in irradiated mice many months after the ovaries have been 

Granulosa-cell tumors first appeared frequently among mice in our 
laboratory after ovaries were transplanted intrasplenically into gonadecto- 
mized mice (11-13). Dr. M. H. Li, who had spent some time trying to 
induce ovarian tumors in fish, first undertook these experiments and his 

Vol. 15, No. 3, December 1954 

702 proceedings: symposium on 25 years of 

Pituitary tumors in estrogen -treated .hybrid mice (C 57 9 *CBA<T) 


<l 2 



42 3 

fc Days of estrogen treatment 


l 65 Days subsequent to estrogen 

i 90.3 treatment 

———————— 120.5 Numbers represent size of tumors 

ih mg. 

— 29.3 








300 400 500 600 700 

Age in days 

Text-figure 7. — Weights of some pituitary glands of estrogen-treated hybrid mice 
killed during the course of prolonged treatment with estrogens or subsequent to 
cessation of estrogen treatment for different times. Evidence of regression sub- 
sequent to discontinuance of treatment was lacking. 

observations were analogous to those previously observed in experiments 
conducted with rats (14) 

We assumed that these tumors were caused by an endocrine imbalance. 
About 25 years ago it was shown that the level of pituitary gonadotrophin 
increases after castration; urinary gonadotrophins increase after the meno- 
pause. Zondek (15) demonstrated that hepatic tissue of rats destroys 
estrogens. Subsequently, many investigators have been concerned with 
quantitative and mechanistic aspects of hepatic inactivation of estrogens. 
Hepatic tissue from mice is similarly active in reducing the biological 
activity of estrogens (16). Because the spleen drains its blood into the 
liver, an ovary located at this site would be unable to have its hormones 
effectively by-pass the liver to reach the systemic circulation. Under 
such conditions the ovary would exist in a physiologically castrated host. 
Such ovaries are stimulated excessively. By use of parabiosis it has been 
demonstrated that the intrasplenic ovarian graft does not prevent aug- 
mented levels of circulating gonadotrophins in the host (17). Some grafts 
and the tumors produced by them are associated with evidence that some 
hormone escapes hepatic inactivation. Absence of a complete castration 
effect is indicated further by the low incidence of adrenal tumors appearing 
in gonadectomized mice with intrasplenic ovaries. 

Journal of the National Cancer Institute 

Pituitary tumors in estrogen treated hybrid mice (C 57 ox C 3 Ho* ) 






- 38.3 










Days of estrogen 

Days subsequent 

117 5 to estrogen 
173 treatment 

Numbers represent 

size of tumors 
in mg. 


- I!!. 3 




400 500 

Age in days 


Text-figure 8. — Weights of some pituitary glands of estrogen-treated hybrid mice 
killed during the course of prolonged treatment with estrogens or subsequent to 
cessation of estrogen treatment for different times. Evidence of regression sub- 
sequent to discontinuance of treatment was lacking. 

Mice of all of the strains that have been studied adequately have ac- 
quired ovarian tumors subsequent to roentgen irradiation or when go- 
nadectomized mice have carried intrasplenic ovarian grafts (12, 13). In 
some strains the tumor may occur at earlier ages than in others but our 
data are not adequate to demonstrate this conclusively. It is the writer's 
impression that mice of strain C57BL acquire fewer ovarian tumors, 
except at great age, than do mice of other strains (C3H, A, CBA, BC). 
The strain differences in the tendency to acquire ovarian tumors are 
not as decisive as with tumors of some other types. Studies undertaken 
up to this time have indicated that tumors in hybrids occur at earlier 
ages than in the parental strains (Gardner, unpublished data) . 

The ovarian tumors were ascribed to hormonal imbalances of excessive 
gonadotrophin; excessive in amount and/or continuity. The incidence of 
ovarian tumors in control mice is very low but they do occur. Dr. Li 
in our laboratory immediately set about checking the hypothesis of a 
humoral etiology of these tumors (13). Under a variety of conditions, 
ovarian tumors did not occur in intrasplenic grafts (text-fig. 9). Ovarian 
grafts in the spleens of intact or of unilaterally gonadectomized mice did 

Vol. 15, No. 3, December 1954 




not become tumorous. Intrasplenic grafts in castrated mice that were 
intentionally or accidentally adherent to the body wall rarely became 
tumorous. The injection of exogenous estrogens or androgens prevented 
tumors in intrasplenic ovarian grafts (text-fig. 10). Ovarian grafts placed 
subcutaneously or intratesticularly seldom became tumorous. All of 
these observations indicate that gonadal hormones prevented granulosa- 




Text-figure 9. — Examples of methods of ovarian grafting in which granulosa-cell 
tumors do not appear. 




Text-figure 10. — Ovarian tumors are inhibited in intrasplenic grafts in castrated 
mice by the injection of sex hormones. 

Journal of the National Cancer Institute 


cell tumors and were compatible with the interpretation of a tumorigenic 
effect of pituitary gonadotrophins (2). 

Two additional types of experiments are highly desirable. First, it 
would be desirable to determine whether hypophysectomy would prevent 
the development of ovarian tumors, and second, to determine whether 
purified gonadotrophic hormones would induce ovarian tumors in hypo- 
physectomized, or for that matter, intact hosts. Attempts in our labora- 
tory to conduct really convincing experiments of these types have been 
unsuccessful up to this time. 

The studies on the influence of hypophysectomy on the late develop- 
ment of ovarian tumors or on ovarian tumorigenesis have been thwarted, 
in part because it has been difficult to keep older mice alive for long 
periods and in part because it is difficult to predetermine tumor growth 
even in intact animals. The natural history of the granulosa-cell tumors 
reveals extreme variability in their progressive growth and behavior 
(Gardner, unpublished data) . Some tumors, after their first appearance, 
remain small for periods of 300 to 400 days; some begin to grow rapidly 
at the end of a long period of inactivity; some grow rapidly immediately 
after they have become apparent; and some grow slowly over periods of 
several hundred days. The variable progression of the tumors in mice 
with intact pituitary glands cannot be explained adequately at this time 
and makes it almost impossible to determine whether hypophysectomy 
would affect the growth of any one tumor or of a small group of tumors. 

What changes occur in the intrasplenic ovarian grafts and are these 
changes different from those in grafts placed in other areas? In all ovarian 
grafts the numbers of ova are depleted more rapidly than in the intact 
organs. Grafts that persist for prolonged periods have bursa-like clefts 
lined by germinal epithelium investing variable portions of their surfaces. 
Ova and follicles may be present for 600 days in grafts placed in sub- 
cutaneous or intramuscular or intratesticular sites but these are rare. 
Ova and follicles are rarely found in intrasplenic grafts in gonadectomized 
hosts after 200 days — they show a precocious ' 'senility.' ' We might 
assume that it is either the precocious aging that is responsible for ovarian 
tumorigenesis or that it is the prolonged exposure to augmented or un- 
interupted gonadotrophins. 

Mice bearing intrasplenic grafts and made hypothyroid by feeding 
thouracil, or hyperthyroid by feeding dessicated thyroid, showed a low 
incidence of tumors in intrasplenic grafts. Inanition also reduced the 
incidence of ovarian tumors in intrasplenic grafts, but they quickly 
appeared when ad libitum feeding was reinstituted (19). 

Dr. Fern Smith in our laboratory a few years ago demonstrated that the 
amount of gonadotrophin in the pituitary glands of intact and castrated 
mice decreased with really advanced age (20). Old mice have atrophic 
ovaries. It is possible that few ovarian tumors occur in old mice because 
the ovaries and pituitary gland decrease in function simultaneoulsy, or 
nearly so. Old ovaries, however, can be transplanted into the spleens of 
young gonadectomized hosts. They do not become tumorous more rapidly 

Vol. 15, No. 3, December 19S4 

316263—54 30 

706 proceedings: SYMPOSIUM ON 25 YEARS of 

than do ovaries of young mice (21, 22). Irrespective of the age of the 
ovary, it seems that the duration of exposure to a modified hormonal 
environment is more important than the persistence of a "senile" ovary. 
The low incidence of "spontaneous" granulosa-cell tumors is probably due 
to the failure of such ovaries to persist for long periods in a relatively 
anovular state and to the decline of pituitary gonadotrophic function with 
advancing age. The combination of precociously senile ovaries in a mouse 
with prolonged effective pituitary function should promote ovarian tumori- 
genesis, unless adrenal-cortical changes supervene. 

When ovaries are depleted and follicles fail to form, ingrowths occur 
from the germinal epithelium and extend into the ovarian stroma. These 
tubular or cleftlike ingrowths may extend throughout and greatly enlarge 
the ovary. These ingrowths may form tumors and are designated tubular 
adenomas when they are large and hyperplastic. Granulosa-cell tumors of 
different histological types also seem to arise from the tubular or cleftlike 
ingrowths of germinal epithelium. Some of the areas of granulosa cells 
may store increased amounts of fatty material and become luteinized — 
luteinized granulosa-cell tumors. Some of the granulosa-cell tumors con- 
sist of follicular structures, some of coarse or fine trabecular structures 
and some of cells that are arranged in masses showing no definite structure. 
The histogenesis of the ovarian tumors of the several types seems thus to 
be quite similar. 

Roentgen rays destroy the ova of mice (10) and mice so treated acquire 
ovarian tumors (9). Germinal epithelial ingrowths again occur in these 
anovular ovaries, ovarian hormonal function is impaired and pituitary 
gonadotrophins increase (9, 24). Estrogens prevent tumorigenesis in such 
ovaries but testosterone propionate (23) does not, although administered 
in amounts that will prevent augmented excretion of gonadotrophin (24) . 
Irradiated ovaries also become tumorous when transplanted into testes or 
when testes are transplanted to the irradiated female mouse (Gardner, 
unpublished data) . With these exceptions the ovarian tumors in irradiated 
mice and mice bearing intrasplenic grafts seem comparable and seem 
to have a similar hormonal etiology. 

Ovarian granulosa-cell tumors grow subsequent to transplantation into 
other animals of the same strain (IS, 25). All that have been transplanted 
have not grown. They grow slowly even when removed and transplanted 
subcutaneously into the donor. Some of the tumors apparently require a 
specific hormonal environment for their rapid growth (26). The period of 
dependency is not too well known and only now are the histologic character- 
istics of dependent and autonomous tumors being studied in detail. Histo- 
logic differences between the dependent and autonomous tumors are not 
apparent at this time. The natural history of the tumors seems most 
significant prognostically; tumors that are growing rapidly, whether 
arising in the younger or the older animals seem most predisposed to grow 
subsequent to transplantation. 

Now we may return to the "tumor" mouse mentioned earlier. How 
can we account for the interesting coexistence of multiple glandular tumors 

Journal of the National Cancer Institute 


and tumors or hyperplasia of responding end organ? Ovarian granulosa- 
cell tumors may develop very slowly; some have been followed after 
original detection of tumorous masses, for almost 400 days. These tumors 
produce estrogens. Only recently have we seen a pituitary tumor in a 
castrated male mouse (A X C57BL) bearing a subcutaneously trans- 
planted irradiated ovary. This male had extremely well developed 
mammary glands although no mammary cancer. The ovarian graft had 
enlarged slowly during 7 months to assume dimension slightly in excess 
of a cubic centimeter. 

The 695-day-old " tumor" mouse of the EI strain presumably acquired 
her ovarian tumors first by prolonged gonadotrophic stimulation of her 
anovular ovaries. She had her last litter 371 days before death. These 
anovular ovaries became tumorous and produced estrogens which in turn 
induced the tumor of the pituitary gland, the latter was responsible for the 
unusual adrenal-cortical proliferation and, with the estrogen, for the 
unusual mammary development. The extent to which genetic influences 
were concerned can also only be surmised or ascertained by analogy as 
this unusual mouse and her daughter, the last of their line, were both 
killed the same day. In other instances, however, the tendency for 
pituitary tumors is strain-specific and presumably this mouse was from a 
strain susceptible to such tumors. 


Approximately 19 years ago one mouse was found that had multiple 
mammary adenocarcinomas, bilateral granulosa-cell tumors and a chromo- 
phobe adenoma of the pituitary gland together with nodular hyperplastic 
adrenal cortices and a hyperplastic endometrium. 

Subsequently by proper combination of strains and hormones, it was 
possible to obtain many pituitary tumors of large size. Estrogen- treated 
mice of the C57BL strain and their hybrids consistently acquired pituitary 
adenomas. Prolonged periods of treatment were necessary. The tumors 
were not reversible in the animal of origin and some grew when trans- 
planted subcutaneously into estrogen- treated mice of the same strain but 
the transplants usually grew only after a prolonged quiescence. 

Ovarian granulosa-cell tumors arising in intrasplenic ovarian grafts in 
castrated mice have appeared in mice of all strains studied and are pre- 
vented when estrogens or androgens are injected or when gonadal hormones 
reach the systemic circulation in appreciable amounts. Intrinsic hor- 
mones, presumably gonadotrophins in this instance, incite ovarian tumors. 

The several types of tumors observed in the one untreated mouse can 
now be obtained at will, by proper combination of transmitted and 
hormonal influence. 

General References 

Gardner, W. U., Pfeiffer, C. A., Trentin, J. J., and Wolstenholme, J. T.: Hor- 
monal factors in experimental carcinogenesis. In Physiopathology of Cancer. 
(Homburger, F. and Fishman, W. H., eds.). New York, Hoeber-Harper, 1953. 

Vol. 15, No. 3, December 1954 

708 proceedings: symposium on 25 years of 

Gardner, W. U. : Hormonal aspects of experimental tumorigenesis (Greenstein, J. P., 
and Haddow, A., eds.). In Adv. Cancer Res. 1: 173-232, 1953. 

Specific References 

(1) Gardner, W. U., Strong, L. C., and Smith, G. M.: An observation of primary- 

tumors of the pituitary, ovaries and mammary glands in a mouse. Am. J. 
Cancer 26: 541-546, 1936. 

(2) Gardner, W. U.: Hormonal imbalances in tumorigenesis. Cancer Res. 8. 

397-411, 1948. 

(3) Trentin, J. J.: Vaginal sensitivity to estrogen as related to mammary tumor 

incidence in mice. Cancer Res. 10: 580-583, 1950. 

(4) : The effect of the presence or absence of the milk factor and of castration 

on mammary response to estrogens in male mice of strains of known mammary 
tumor incidence. Cancer Res. 11: 286-287, 1951. 

(5) Gardner, W. U., and Strong, L. C: The strain-limited development of tumors 

of the pituitary gland in mice receiving estrogens. Yale J. Biol. & Med. 12: 
543-548, 1940. 

(6) Gardner, W. U.: The effect of estrogen on the incidence of mammary and 

pituitary tumors in hybrid mice. Cancer Res. 1: 345-358, 1941. 

(7) : Hormones in experimental carcinogenesis. Acta Unio Contra Cancrum 

6: 124-133, 1948. 

(8) Furth, J.: Thyroid-pituitary tumorigenesis. J. Nat. Cancer Inst. 15: 687-691, 


(9) Furth, J., and Butterworth, J. S.: Neoplastic diseases occurring among mice 

subjected to general irradiation with X-rays. II. Ovarian tumors and associated 
lesions. Am. J. Cancer 28: 66-95, 1936. 

(10) Brambell, F. W. R., and Parkes, A. S.: Changes in the ovary of the mouse fol- 

lowing exposure to X-rays. III. Irradiation of the non parous adult. Proc. 
Roy. Soc, London, s.B, 101: 316-328, 1927. 

(11) Li, M. H., and Gardner, W. U.: Tumors in intrasplenic ovarian transplants in 

castrated mice, Science 105: 13-15, 1947. 

(12) : Experimental studies on the pathogenesis and histogenesis of ovarian 

tumors in mice. Cancer Res. 7: 549-566, 1947. 

(IS) : Further studies on the pathogenesis of ovarian tumors in mice. Cancer 

Res. 9: 35-41, 1949. 

(14) Biskind, M. S., and Biskind, G. S.: Development of tumors in the rat ovary 

after transplantation into the spleen. Proc. Soc. Exper. Biol. & Med. 55: 
176-179, 1944. 

(15) Zondek, B. : tTber das Schicksal des Follikelhormons (Follikulin) im Organismus. 

Skandinav. Arch. f. Physiol. 70: 133-167, 1934. 

(16) Rush, B. Jr.: Inactivation of estradiol by the hepatic tissues of mice. Proc. Soc. 

Exper. Biol. & Med. 74: 712-714, 1950. 

(17) Miller, O. J., and Pfeiffer, C. A.: Demonstration of increased gonadotrophic 

hormone production in castrated mice with intrasplenic ovarian grafts. Proc. 
Soc. Exper. Biol. & Med. 75: 178-181, 1950. 

(18) Gardner, W. U.: The effect of ovarian hormones and ovarian grafts upon the 

mammary glands of male mice. Endocrinology 19: 656-667, 1935. 

(19) Miller, O. J., and Gardner, W. U. : The role of thyroid function and food intake 

in experimental ovarian tumorigenesis in mice. Cancer Res. 14: 220-226, 

(20) Smith, F. W., and Gardner, W. U.: Biological assay of mouse pituitary gonado- 

trophin. (Abstract.) Anat. Rec. 106: 248, 1950. 

(21) Li, M. H., and Gardner, W. U.: Influence of age of host and ovaries on tumori- 

genesis in intrasplenic and intrapancreatic ovarian grafts. Cancer Res. 10: 
162-165, 1950. 

Journal of the National Caneer Institute 


(22) Klein, M.: Ovarian tumorigenesis following intrasplenic transplantation of 

ovaries from weanling, young adult, and senile mice. J. Nat. Cancer Inst. 12: 
877-881, 1952. 

(23) Gardner, W. U.: Ovarian and lymphoid tumors in female mice subsequent to 

roentgen-ray irradiation and hormone treatment. Proc. Soc. Exper. Biol. & 
Med. 75: 434-436, 1950. 

(24) Chang, C. H., and Van Eck, G. J.: Action of testosterone propionate on the 

pituitary activity of castrated or X-rayed female mice in parabiosis with normal 
females. (Abstract.) Cancer Res. 12: 254, 1952. 

(25) Bali, T., and Furth, J.: Morphological and biological characteristics of X-ray 

induced transplantable ovarian tumors. Cancer Res. 9: 449-472, 1949. 

(26) Cliffton, E. E., and Pan, S. C: The effect of progesterone compound on growth 

of a transplanted granulosa cell tumor. Proc. Soc. Exper. Biol. & Med. 69: 
516-518, 1948. 

Vol. 15, No. 3, December 1954 

Induced Ovarian and Adrenal 

Tumors 1 ' 2,3 

Kathakine P. Hummel, Roscoe B. Jackson 
Memorial Laboratory, Bar Harbor, Maine 

If ovaries of strain DBA mice are subjected to hormonal imbalance by 
grafting to spleens of gonadectomized hosts they become tumorous 7 
months later. This is longer than the process takes in mice of some other 
strains as, for example, in strain A where the ovaries become tumorous 
in 3 months, and in strains C57BL and BALB/c where the process takes 
5 months. 

The normal aging changes in ovaries of strain DBA mice have been 
described by Fekete (J?). A characteristic feature of these ovaries is the 
retention and hyalinization of corpora lutea. Tumorous changes are 
rarely found. In a recent survey, one tumor, a tubular adenoma, was 
found among 13 DBA mice over 20 months of age. 

To determine how long an ovary must be subjected to hormonal im- 
balance before being irreversibly altered, DBA ovaries were grafted to 
spleens of gonadectomized hosts, left there for periods of time ranging 
from 3 days to 5 months and then removed and transferred to the normal 
site or ovarian bursa of a second host, this time a hybrid DBA X C3H. 
In order that the hormonal environment of this host might be as normal 
as possible only one ovary was replaced, the other being removed 4 to 6 
days later after the grafted ovary had established itself and could function 
normally. The hybrids bearing grafted pre treated ovaries were mated 
and in many cases produced normal young (2). As has been generally 
observed with transplanted ovaries, considerably fewer litters were pro- 
duced by graft-bearing mice than by unilateral castrates or intact mice. 
The reproductive function of ovaries previously resident in spleens for 
periods up to 1 month did not differ essentially from ovaries transplanted 
directly without splenic sojourn. Reproductive function, however, did 
decrease with the length of time the ovary had remained in the spleen, no 
young being produced from ovaries that had been in spleens for over 2 
months. There was no similar decrease in hormonal function and no 
differences were noted in matings as evidenced by vaginal plugs nor in 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 29, 1954. 

2 This is a report of several years' work involving a group of investigators, notably Mary S. Eddy and Barbara 
Rupple, who were responsible for the grafting operations in large part and Dr. Elizabeth Fekete, who kindly 
assisted in the tumor diagnoses. 

3 This work has been supported in part by a grant-in-aid from the American Cancer Society upon recommenda- 
tion of the Committee on Growth of the National Research Council, and by a research grant C1073(C) from the 
National Cancer Institute of the National Institutes of Health, U. S. Public Health Service. 


Journal of the National Cancer Institute. Vol. 15, No. 3, December 1954 


712 proceedings: SYMPOSIUM ON 25 YEARS OF 

stimulation of mammary glands as evidenced by the presence of mammary 
gland tumors. In fact, mammary gland tumors occurred in the hybrid 
graft-bearing mice in such numbers, and at such an early age, that it was 
necessary to use hybrids deprived of the milk agent in later experiments. 

In 97 of 159 animals (61%) killed 12 to 24 months after receiving the 
graft, the ovaries were found to be tumorous, 31 of these being grossly 
observable. Sixteen of 134 animals (12%) killed earlier than 12 months 
following grafting had tumorous ovaries, all of these being microscopic 
in size. In the control transplantation animals (hybrids grafted with 
DBA ovaries not previously resident in spleens), 5 of 9 (or 55%) killed 12 
to 24 months after the transplantation had tumors of the ovaries. No 
tumors were observed in 10 controls killed earlier than 12 months. 

We concluded from these results that the hormonal imbalance imposed 
upon the ovaries by grafting them into the spleens of gonadectomized 
hosts has little to do with the incidence of ovarian tumors and that trans- 
plantation per se must be an important factor. The period of splenic 
sojourn apparently hastens tumorigenesis, as ovarian tumors were found 
earlier if the period of imbalance had lasted at least 2 weeks. Ovarian 
tumors were found 8 months after ovarian transplantation in these cases, 
whereas they were not found until 14 months later if the sojourn in spleen 
was of 3 or 7 days' duration. Tumors of the ovary were found 15 months 
after direct transplantation in the control series. 

Some of the factors that may have been operative in the induction of 
these ovarian tumors following transplantation are briefly summarized. 

1) Hybrid environment. — Although no tumors have been seen in ovaries 
of intact or unilaterally castrate hybrid DBA X C3H mice, it is well 
known that the hybrid environment often gives rise to abnormalities not 
seen in either parent. 

2) Transplantation and consequent breakdown of a large part oj the ovarian 
tissue. — The ovary becomes reconstituted, but never completely — as is 
evidenced by an early decrease in the number of ripening follicles and the 
short reproductive life. The small amount of functional ovarian tissue 
may create an abnormal ovary-pituitary hormone balance. 

3) Abnormal proportions oj pituitary gonadotrophins released. — The 
remnant or fragment of ovary undergoes compensatory hypertrophy 
partly through accumulation of interstitial cells. These have been shown 
histochemically to store and possibly secrete the precursor of gonadal 
hormones (cholesterol). Mobilization and release of gonadal hormones 
is dependent upon the proportion of the gonadotrophins F.S.H., L.H. and 
L.T.H. released by the pituitary. In these ovaries there is an over- 
growth of interstitial cells suggesting increased precursor storage, and in 
turn disproportionate gonadotrophin release from the pituitary. 

4) Disproportionate release of gonadal hormones. — A hypertrophied 
interstitium is often associated with increased androgen production by 
an ovary. The conspicuous interstitium of these ovaries suggests imbal- 
ance within the ovary in the estrogen, androgen and progesterone produc- 
tion and/or release, and between precursor and hormone. 

Journal of the National Cancer Institute 



Adrenal glands of many of the hybrid graft-bearing mice were saved 
and examined. In the control series adrenals from 14 animals were 
studied. Of these, one in an animal with an ovarian tumor killed 13 
months after transplantation showed slight nodular hyperplasia of the 
cortex. None of the others showed tumorous changes in the adrenals. 
In the experimental series (hybrids with grafted ovaries previously sub- 
jected to the hormonal imbalance) many of the adrenals showed cortical 
nodular hyperplasias and carcinomas characteristic of castrate mice. 

Adrenals of 223 hybrids bearing grafted ovaries that had been previ- 
ously resident in spleens were examined. In 46, the ovarian graft was 
absent or minute and in these, as might be expected, 42 or 91 percent had 
adrenal-cortical tumors. Table 1 summarizes the observations in the 
remaining 177 hybrids. In 33, or 19 percent, killed at an average age 
after graft operation of 9 months there were neither adrenal nor ovarian 
tumors. One third of the animals had both adrenal and ovarian tumors, 
42 (or 24 percent) had ovarian tumors but not adrenal tumors, and 46 
(or 26 percent) had adrenal but not ovarian tumors. There is a suggestion 
that the adrenal tumor formation preceded the ovarian in that they were 
found in animals killed at an average of 15 months after grafting of the 
ovaries, whereas ovarian tumors were found at an average postoperational 
age of 18 months. It is noteworthy that the characteristic retention and 
hyalinization of corpora lutea was not found in any ovary of strain DBA 
grafted to a hybrid. Seventy-five percent of the hybrids with both ovarian 
and adrenal tumors had had functional ovaries in that they had either 
borne young or had vaginal plugs as evidence of estrus and mating. 

Table 1. — Ovarian and adrenal tumors in 177 (DBAXC3H) hybrids with grafted DBA 
ovaries "altered" by sojourn of 8 to 5 months in spleens of castrates 

of mice 



With functional ovaries 





Type of 








Ovarian + 
Adrenal + 

Ovarian + 
Adrenal — 

Adrenal + 


56 (32%) 
42 (24%) 
46 (26%) 
33 (19%) 









42 (75%) 



14 (25%) 

This finding of adrenal and ovarian tumors in the same experimental 
animal indicates causation by a single type of hormonal imbalance. This 
supports the theory held by some that the basic mechanism of hormonal 
imbalance is the same in ovarian tumors induced by X ray, ovarian tumors 

Vol. 15, No. 3, December 1954 

714 proceedings: SYMPOSIUM ON 25 YEARS OF 

induced by implantation to spleens of gonadectomized mice, and adrenal 
tumors induced by neonatal castration. The gonad-pituitary balance is 
upset. One widely held theory is that a pituitary released from gonadal- 
hormone inhibition releases gonadotrophins of abnormal quality, or in 
excessive amounts over a prolonged period of time. 

It would seem in the case of our experimental animals that the ovarian 
and adrenal-cortical tumors have resulted from an hormonal imbalance im- 
posed upon the pituitary-gonad-adrenal complex by a fragment or remnant 
of functional ovarian tissue. In support of this theory, preliminary ob- 
servations on a group of DBA X C3H hybrids in which gonadectomy was 
incomplete (a small remnant of ovary having been left) show adrenal- 
cortical tumors one year later — even though the remnant of ovary func- 
tioned to produce young. Complete gonadectomy at 2 months of age, 
approximately the age at which the ovary grafts were made, results in 
adrenal tumors 12 months later. No ovarian or adrenal tumors have been 
found in intact or unilaterally castrate DBA X C3H hybrids up to 24 
months of age. 

Morphologically the abnormal adrenals resemble the abnormal ovaries. 
Large accumulations or nodules of pale hypertrophied cells which are 
similar in appearance to the interstitial cells of the ovary are present. 
There are masses of lipochrome cells that in the ovary are remnants of 
atretic follicles. There are cysts among the hypertrophied cells into which 
clumps of cells penetrate. In such clumps, the granulosa-cell tumors of the 
ovary seem to arise and in similar areas adrenal-cortical carcinomas appear. 
The adrenal tumors often assume a folliculoid structure difficult to dis- 
tinguish from ovarian tissue. Also in these adrenals, cysts lined with 
ciliated epithelium, similar to cysts found in transplanted DBA ovaries, are 
seen. Such cysts are frequently found at the hilus of the ovary in some 
strains of mice, notably C57L (8) where they are thought to be remnants 
of the mesonephros. The presence and origin of these cysts in the adrenal 
cortex is puzzling as they have been seen only in DBA X C3H mice bearing 
grafted DBA ovaries. The morphologic picture of the adrenal cortex 
leads us to believe that it functions hormonally as a second ovary under 
the influence of gonadotrophins from a pituitary that, in turn, is responding 
to a lack of normal circulating gonadal hormones. 

In summary, our experiments have shown that transplantation of ovaries 
of DBA mice into DBA X C3H hybrids results in tumorous changes in the 
grafted ovaries and also in the cortex of the adrenal of the host. This is 
due to an upset in the ovary-pituitary-adrenal complex possibly occasioned 
by deficiency of or abnormal qualities of gonadal hormones from an ovary 
partly destroyed by transplantation. After transplantation the ovary 
does not fully recover its reproductive or endocrine function. This 
creates an abnormal and irreversible ovary-pituitary balance. Compen- 
satory mechanisms involving the adrenal cortex set up a transitory balance 
that can not be maintained for any length of time and abnormal growth 
of ovary and adrenal cortex is the end result. 

Journal of the National Cancer Institute 



(1) Fekete, E.: A comparative study of the ovaries of virgin mice of the dba and 

C57 black strains. Cancer Res. 6: 263-269, 1946. 
{2) Little, C. C, Hummel, K. P., Eddy, M. S., and Rupple, B.: Young produced 

from ovaries subjected to endocrine imbalance. Proc. Nat. Acad. Sc. 37: 

666-669, 1951. 
(5) Fekete, E.: A morphological study of the ovaries of virgin mice of eight inbred 

strains showing quantitative differences in their hormone producing components. 

Anat. Rec. 117: 93-113, 1953. 

Vol. 15, No. 3, December 1954 



Carcinogenesis in the Adrenal *• 

George W. Woolley, Chief, Division of Steroid 
Biology, Sloan-Kettering Institute for Cancer Re- 
search, and Professor of Biology, Sloan-Kettering 
Division, Cornell University Medical School, New 

I shall take the adrenal gland as an example for my discussion. I am 
glad to do this as it involves a site that has long been of particular interest 
to me. I can well remember standing at a table in the old Jackson Lab- 
oratory, autopsying a dilute brown mouse which had been castrated at 
birth and now aged and with breast cancer, and noting for the first time 
adrenal changes later described as nodular hyperplasia of the adrenal. 
Through transplantation experiments the ovary-like endocrine secretion 
present in this and other similar ovariectomized animals was determined 
to have arisen from the adrenal glands. 

I was particularly interested in the condition of the adrenal since as a 
graduate student at Wisconsin I had heard discussions on endocrines refer- 
ring to the adrenal cortex as "the great unknown" in endocrinology. 
How far knowledge has progressed since that time! The adrenal is now 
one of the best known glands. 

Subsequently, a histological interpretation of the adrenal-cortical 
changes was developed with the aid of Dr. Elizabeth Fekete and Dr. A. 
M. Cloudman, of the Jackson Laboratory. 

Adrenal-cortical tumors similar to those of the mouse have now been 
described in such experimental animal forms as the guinea pig, the rat, 
and the hamster. 

To reminisce a little further — ideas regarding etiological, or origin rela- 
tionships of adrenal-cortical tumors go back to the last century. In 1891, 
Marchand described a case where a human female hermaphrodite had, on 
autopsy, atrophied ovaries and greatly enlarged adrenals. In a similar 
case, Creccio found that the adrenals had increased to the size of the 

An experimental test of the idea of ovary-adrenal relationship appears 
to have first been performed by Miss H. E. Feodossiew, at the University 
of Kazan. The results of this test were published in 1906. The experi- 
ment seems to have eluded a number of workers in this country. 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine 
June 29, 1954. 

8 This investigation was supported in part by a research grant (C-1796) from the National Cancer Institute, 
of the National Institutes of Health, U. S. Public Health Service, and in part by a grant from the American Cancer 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

718 proceedings: SYMPOSIUM ON 25 YEARS OF 

One can find that Professor Lubimov and his student, Miss Feodossiew, 
clearly understood that a relationship existed between the ovary and the 
adrenal cortex. In exploring this relationship they used dogs. These were 
castrated and then autopsied at definite intervals of months. The adrenal 
glands were studied and adrenal-cortical hyperplasia was described. The 
hyperplasia started in the region of the glomerulosa and progressed to the 
fasciculata. The hyperplasia in both layers was focal, i.e., in islands. 
The growth reached and penetrated the medulla in one direction and 
progressed through the capsule in the other, forming in the latter case a 
mushroom-like growth which remained united to the cells of the peripheral 
layer. Increase in the number of mitotic figures was noted. All of this 
sounds remarkably modern. 

During the past few years we have come to understand endocrine rela- 
tionships more clearly, and to differentiate these relationships from other 
important etiologic factors — genetic and vascular. 

The first of these factors to be mentioned is the genetic one — the in- 
fluence of heredity. 

We know that this influence exists in the mouse because of the presence 
and persistance of strain variations in the occurrence of the adrenal-cortical 
tumors. Some strains have adrenal-cortical tumors without gonadectomy, 
others only with gonadectomy. Some strains have, characteristically, 
only a few or no cortical tumors; others, benign tumors; and still others, 
malignant tumors. Tumors of certain strains produce, characteristically, 
estrogenic hormones; others, androgenic hormones; and some produce both. 
A few, especially late in life, give evidence of producing progesterone-like 

It is not known whether the genetic factors act most effectively system- 
ically through hormone imbalance, or locally in the adrenal-cortical tissue. 
The evidence of Huseby and, thus far, our own evidence, indicates that 
an important site of action is in the adrenal-cortical tissue itself. 

Heterosis is a modifying factor and acts in a positive direction toward 
tumor-formation — as far as it has been studied. 

Carcinoma tends to be dominant over hyperplasia, and hyperplasia to 
be dominant over no tumor in crosses between strains with different tumor 

The second influence to be considered in more detail is that of the 
endocrine secretions. 

The occurrence of adrenal-cortical tumors may be retarded by removal 
of the pituitary, the use of steroid hormones related to those of the adrenal, 
ovary, or testes, and by the presence of hormone-producing tumors. 
Both thiouracil and caloric restriction retard hormone production by the 
tumors. The latter influence has been interpreted as being similar to 
partial hypophysectomy. 

Occurrence of adrenal-cortical tumors may be enhanced by gonadectomy, 
by excess estrogen, by pituitary hormones, and by transplantation of 
ovaries to the spleen. In the latter case, it is thought that enhancement 

Journal of the National Cancer Institute 


occurs because estrogenic hormones are inactivated in the liver before 
reaching the pituitary. 

A third influence affecting carcinogenesis, and the latest to be found, 
appears to be a mechanical one and related either to nutrition or to 
vascularization of the adrenal. The influence of this factor has been 
demonstrated by transplantation of the adrenal to a subcutaneous site. 
It is a tumor-retarding influence. 

Finally, the occurrence of adrenal-cortical tumors requires the integra- 
tion of genetic, hormonal, and site factors — and possibly many more as 
yet undiscovered factors. Carcinogenesis in the adrenal appears to be 
a complex syndrome but one capable of progressive solution and under- 

A possible theory of origin might be the following: in the absence, or 
deficiency, of gonadal hormones — ovarian or testicular — a normal and 
proper hormonal brake is not applied to the pituitary. The excess pitui- 
tary hormones which then accumulate act on the adrenal cortex, a tissue 
arising embryologically from the sex ridge. If this latter tissue has 
genetic factors for tumor susceptibility, and proper vascularization and 
nutrition, tumors will develop. 

Vol. 13, No. 3, December 1954 

Discussion: Carcinogenesis in Endocrine Organs 

Dr. William B. Atkinson, University of Cincinnati School of Medicine, 

Cincinnati, Ohio 

Among the many approaches to the study of the pathogenesis of neoplasms, few 
have been as productive as that of the investigation of abnormal growth in endocrine 
organs. It may be somewhat trite, but nonetheless valuable, to remind this group 
that one of the most venerable definitions of neoplasia is that of uncontrolled growth 
of a tissue. In this very definition, however, we find the clue to the greatest obstacle 
in the elucidation of carcinogenesis; i.e., that before we can deal successfully with the 
problem of "uncontrolled" growth, we must first know at least the basic physiologic 
processes involved in the normal growth and differentiation of the tissues composing 
the various organs of the body. We must learn to manipulate normal growth in 
order to establish techniques for the investigation of abnormal growth. In most 
tissues and organs of the body our present knowledge, unfortunately, is entirely 
inadequate in this respect. With all deference to the excellent work reported previ- 
ously in this conference, insofar as carcinogenesis is concerned, it is of little comfort or 
ultimate edification to know simply that the development of a particular tumor may 
be associated with the action of a gene or group of genes. Likewise, it helps us but little 
to record the effects of X radiation or chemical carcinogens upon the skin, for instance, 
until we have a great deal more fundamental information concerning this organ of 
the kind reported by Dr. Chase in last evening's session. 

It must be admitted that the endocrinologist has enjoyed several unique advantages 
in his study of tissue growth. First, endocrine organs and target organs depend to 
a considerable extent upon known humoral agents for their morphologic and func- 
tional maturation; and second, normal growth and differentiation in the endocrine 
system has been studied with ever-increasing intensity over the last half century. 
Another factor of prime importance has been that, in many instances, the clinical 
conditions with which the physician has had to deal have their closely parallel counter- 
parts in experimental animals. We might cite the virilizing adrenal carcinomas, 
adenocarcinoma of the mammary gland and the uterine hyperplasias, to name but 
several. This has led to the ready exchange of data, and, more important, ideas that 
have often been of immediate usefulness in both the laboratory and clinic. 

With respect to the several papers presented this morning, I have been intrigued 
by two questions which are quite general in scope: 1) with the principal exception of 
the mammary gland, experimentally induced and even "spontaneous" {i.e., genetically 
induced) tumors of the endocrine system are found to occur mainly in the hormone- 
producing organs themselves (such as the pituitary gland, the gonads, the thyroid, or 
the adrenals) . The consistent production of neoplasms in the final target organs (such 
as the uterus, seminal vesicles, submaxillary glands or kidneys) is far less commonly 
obtained. Why? 2) Dr. Furth, in particular, has reminded us of the increasing 
autonomy of individual tumors with respect to their environment; that neoplasms 
which may survive only autotransplantation in their early stages of development, 
may eventually become transplantable even into unrelated species. I believe that 
this important biological property of neoplasms is one which must eventually be 
taken into greater account in studies on histocompatibility. I am sure we should 
all like to hear any comments that today's speakers may care to make on the fascinat- 
ing question of why the growth of a tumor transplanted in its dependent phase of 
development may be blocked entirely by one or more "incompatible" genes in the 
host, whereas the same tumor transplanted in its later autonomous phase may grow 
with complete disregard of what was earlier a hostile genetic environment. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 
316263—54 31 

722 proceedings: symposium on 25 years of 

Dr. Flavia L. Richardson, Roscoe B. Jackson Memorial Laboratory, 
Bar Harbor, Maine 

We have studied the effect of castration and adrenalectomy on the development of 
the mammary glands of female mice. Female mice of strains BALB/c and C57BR/cd 
were castrated 3 days after birth. These mice were sacrificed at 3 months of age and 
the mammary glands were compared with those of normal females of the same age. 
There was very little development of the mammae in the castrated BALB/c mice. 
The ducts were very narrow, had few branches, and covered a much smaller area 
than in the normal females. In contrast, the mammary glands of the castrated 
C57BR/cd females showed considerable variation ranging all the way from small 
glands similar to those in the BALB/c castrates to glands that very closely resembled 
those of the normal C57BR/cd females. The ducts were wide, well branched and 
covered a large area. The uteri of all castrated animals showed no stimulation. 
None of the . castrated animals possessed the numerous lateral buds or very small 
acini that were present in some of the glands of normal animals. 

C57BR/cd females, castrated at 3 days of age, were adrenalectomized when about 
25 days old. The mammary glands of animals 3 months of age closely resembled 
those of the mice that were castrated only. 

Dr. Robert S. Speirs, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

Dr. Gardner has clearly illustrated how experimental manipulations of genetical 
and hormonal factors can produce growths that are identical to spontaneously occurring 
tumors. His excellent presentation brought out a number of factors that I would 
like to discuss briefly. 

The effects of hormones and other substances upon tumor cells are usually deter- 
mined only by gross measurements of weight and size. These are acknowledged 
to be crude, and exceedingly variable from mouse to mouse, and are more qualitative 
than quantitative. This is especially true of tumors that are transplanted and grown 
subcutaneously. However, in recent years it has been possible to adapt many different 
tumors for growth in the peritoneal cavity (1). In the peritoneal cavity, the tumor 
cells are suspended in a fluid medium, thus permitting the application of modified 
hematologic techniques. In this manner, the total number of cells per cubic millimeter 
can be determined by the chamber method, stained smears can be obtained for differ- 
ential counting and estimates can be made of the total amount of fluid in the peritoneal 
cavity. Thus the total number of cells of each component of a tumor, as well as the 
local cellular response of the host to the tumor, can be determined. 

The subcutaneous air pouch, developed by Dr. Hans Selye (2), is an artificial sac, 
which is somewhat more accessible than the body cavities but is very similar in other 
respects. Gross measurements of fluid volume as well as quantitative measurements 
of cells can be performed daily or even hourly if necessary. Moreover, the whole 
sac containing the tumor cells can be surgically removed at any time, and if desired 
a second sac can be produced for fresh tumor cells. 

This procedure of growing the tumor cells as a suspension within the peritoneal 
cavity or pouch, permits a quantitative determination of the growth rate of each type 
of cellular component of a tumor under normal and experimental conditions. This 
method of measurement should be more accurate and sensitive to experimental 
procedures than the gross measurements of size and weight. 

It is important to determine whether the hormones act directly upon tumor cells, 
or whether they affect the defense reactions of the host and thereby modify tumori- 
genesis and growth by indirect means. This type of problem can easily be studied 
in the tumors adapted for growth in the peritoneal cavity or pouch. It is commonly 
known that substances injected into an inflamed peritoneal cavity or air pouch, do 
not pass into the circulation very readily. It is therefore possible to study the effects 
of a particular substance acting locally upon the tumor cells, and compare it with the 

Journal of the National Cancer Institute 


effects of the same substance given systemically. Moreover, because these tumor 
cells are growing in a physiologically isolated type of tissue culture, it is possible to 
utilize the peritoneal cavity or air-pouch for culturing viruses or even for growing 
certain tissues such as the pituitary or spleen together with the tumor cells. 

The final technique which I would like to comment upon is that of hypophysectomy. 
In the clinic there is a great deal of interest in the effect of this operation upon patients 
with inoperable cancer and in some cases it certainly seems to have a striking effect. 

The effect of hypophysectomy upon tumorigenesis and growth should certainly 
be investigated more thoroughly, especially in the mouse where a wide variety of 
tumors are readily available for study. In the mouse, hypophysectomy is somewhat 
difficult to learn; however, once mastered it is relatively easy and quickly performed. 
Using a modification of the technique of Thomas (S) we have found that the entire 
operation can be performed in less than 10 minutes in young mice and that the animals 
may live up to a year or more postoperatively. Furthermore, very young mice 
weighing as little as 6 grams can be hypophysectomized. A number of laboratories 
have utilized hypophysectomized mice for various studies. In 1946, Korteweg and 
Thomas reported on the outcome of 351 hypophysectomies (4) and Dr. Gardner has 
recently stated that well over 1,000 such operations have been performed at Yale. 

The evidence at the present time seems to be that hypophysectomy reduces or 
prevents the formation of mammary cancer in certain stocks of mice with a high 
predisposition to that type of cancer. This is not too surprising in as much as the 
pituitary does directly or indirectly regulate the growth of the mammary tissue. 
However, it would be extremely interesting to know if tumors in other tissues would 
also be inhibited or delayed by hypophysectomy. The rate of growth of transplantable 
or spontaneous tumors can also be studied in hypophysectomized animals. In this 
way one can approach the relation of the pituitary secretions to tumorigenesis and 
tumor growth. 


(1) Hauschka, T. S. : Cell population studies of mouse ascites tumors. Trans. New 

York Acad. Sc. 16: 64, 1954. 

(2) Selye, H.: The diseases of adaption: Rec. Prog. Hormone Res. 8: 117, 1953. 

(5) Thomas, F.: A technique for hypophysectomy of the mouse. Endocrinology 23: 

99-102, 1938. 
(4) Korteweg, R., and Thomas F. : Hypophysectomy in mice with special reference to 

mammary cancer. Cancer Res. 8: 385-395, 1946. 

Vol. 15, No. 3, December 1954 


Session VI . Genetic Control of Behavior 

Chairman, Dr. Frank A. Beach, Department of 
Psychology, Yale University, New Haven, Conn.; 
Board of Scientific Directors, Roscoe B. Jackson 
Memorial Laboratory 

Speaker: Dr. Curt P. Richter 

The Effects of Domestication and Selection on the Behavior of the Norway 

Speaker: Dr. John Paul Scott 

The Effects of Selection and Domestication Upon the Behavior of the Dog 

Speaker: Dr. Laurence H. Snyder 

The Effects of Selection and Domestication on Man 

Discusser, all papers: Dr. John L. Fuller 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

The Effects of Domestication and Selec- 
tion on the Behavior of the Norway 
Rat h 2 

Curt P. Richter, Ph.D., Psychobiological 
Laboratory, Johns Hopkins School of Medicine, 
Baltimore, Md. 

The Norway rat may be considered to be the first animal to have become 
domesticated for strictly scientific purposes. It offers excellent oppor- 
tunities for studying the anatomical, physiological and behavioral changes 
of domestication for the following reasons: 1) The live wild Norway rat is 
readily available in large numbers throughout the world in cities, towns, 
and on farms, and equally large numbers of domesticated Norway rats are 
available in scientific laboratories throughout the world. 2) Since the 
domesticated Norway rat has been used in almost every field of biological 
research, more is known about it than any other animal, with the possible 
exception of man. 3) Wild and domesticated Norway rats breed with 
one another. Neither breeds, as far as is known, with any other rat — 
not even the Alexandrine or roof rat, which next to the Norway is the most 
common rat in the world. The fact that the Norway and Alexandrine 
rats do not breed is extraordinary in view of their great similarity in 
appearance, only the initiated being able to tell them apart. 4) The short 
life span of this rat, and early age of maturity make it possible to follow the 
inheritance of various characteristics throughout many generations within 
a few years' time. 5) The Norway rat is very similar to man in many ways, 
particularly in dietary needs, geographic distribution, world population 
and colony formation. 

Historical Background 

Much of our knowledge of the history of the Norway rat, Mus norvegicus, 
comes from the painstaking search of the literature by H. H. Donaldson, 
who more than anyone else is responsible for the popularity of the Norway 
rat for scientific research. His book The Rat gives an excellent account of 
the history of this animal and many hundred references (1). 

The wild Norway rat did not reach Europe until 1730 and America 
until sometime later, near the end of the 18th century. In Europe, up 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 29, 1954. 

2 This work was carried out in part under contracts between the Office of The Surgeon General, Department 
of the Army, and the Office of Naval Research, Department of the Navy, and the Johns Hopkins University; 
and also in part under a grant between the U. S. Public Health Service and the Johns Hopkins University. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

728 proceedings: symposium on 25 years or 

until that time, another rat, Mus rattus, otherwise known as the Alexan- 
drine, roof, or ship rat, had for many centuries been the most common 
rat. In America this rat, not a native, was well settled at the time of the 
Norway rat's arrival. In all instances the Norway, the fiercer, more 
aggressive rat, soon largely replaced the less virile Alexandrine rat. 

The Norway rat originally came to Europe from somewhere north of 
India by trade routes and ships. Once ships became equipped with guards 
for hawsers this means of transportation was more or less ended for the 
heavy clumsy Norway, but still open for the light nimble Alexandrine rat. 
At the present time many ships still carry rats, but only the Alexandrine 

When and how Norway rats were first domesticated is not known. All 
we can do at this time is to venture a few guesses on the basis of experience 
and knowledge derived, in part, from attempts to tame trapped wild rats 
and first-generation wild rats. 

Since there is no evidence of the Norway rats' having made their ap- 
pearance in captivity before their invasion of Europe, the process of 
domestication must have started sometime after their first arrival in 
Europe in 1730. 

They did not make their appearance in scientific laboratories until much 
later. In 1856, Waller and Philipeaux (2) used them for experiments on 
the effects of adrenalectomy, which curiously is still one of their chief 
uses at the present time. This is the earliest record that we have been 
able to find, so far, of their use for scientific purposes. These authors 
referred to the albinos as Mus rattus, but it is more likely that these rats 
were of the Norway strain, since there was a wrong usage of the names at 
that time. 

Long before Philipeaux, the great French physiologist, Magendieis 
reputed to have recommended the use of rats for physiological experiments. 
So far, however, we have not been able to find any record that he himself 
actually used them. 

After that, they were used quite sporadically by various workers in 
Europe, and also in America, until the time of the founding of the Wistar 
Institute in 1907 and the establishment of the first standard strain of 
albino rats. Donaldson was first introduced to them in 1893 by my former 
chief Adolf Meyer (8). Of the many strains of Norway rats now known, 
the Wistar is most widely used. The source of the domesticated rats 
used in this country remains unknown, that is, we don't know whether 
they came from this country or from Europe. 

In what form and how the Norway rat first made its appearance in the 
laboratory is not known. For several reasons we believe it came in its 
albino form. The rats that Waller and Philipeaux used were albinos, 
and with few exceptions, present-day domesticated Norway rats all over 
the world are albinos. Their clean white appearance has undoubtedly 
had much to do with their popularity. Defects in vision, owing to a 
lack of pigmentation, may have tended to make them less apt to escape 
and easier to handle. 

Journal of the National Cancer Institute 


It is also quite likely that Norway rats came into captivity as albinos. 
We know that rat-baiting was popular in France and England as early as 
1800, and in America soon afterwards. This sport flourished for seventy 
years or more, until it finally was stopped by decree. In this sport 100 to 
200 recently trapped wild Norways were placed at one time in a fighting 
pit. A trained terrier was put into the pit. A keeper measured the time 
until the last rat was killed. Sportsmen bet on the killing times of their 
favorite terriers. For this sport many Norway rats had to be trapped and 
held in pounds in readiness for contests. Records indicate that albinos 
were removed from such pounds and kept for show purposes and/or 
breeding. It is thus very likely that these show rats, that probably had 
been tamed by frequent handling, found their way at one time or another 
into laboratories. That albino rats do appear with some frequency is 
known from the fact that Hatai obtained albinos in 2 of 6 third-generation 
litters of wild Norway rats that were bred in his laboratory (4). 

In our experience wild Norway rats, when trapped while young, can 
be trained by constant handling so that they can be held in the hand 
without danger. Although obviously highly nervous, they remain tame 
as long as they are handled and well fed. When, however, they are not 
handled and are left to themselves, they generally become wild and 
unmanageable and make use of the first opportunity to escape. For 
this reason, and since children quickly tire of them, rats have never made 
good pets. 

Once Norway rats were brought into laboratories their obvious virtues 
for scientific work over the then popular rabbit must quickly have become 
apparent. From generation to generation they became more tractable, 
less apt to escape (as will be explained below) and so, soon changed from 
their role as captive wild rats to domesticated rats. 

The domesticated Norway rat's advantages for scientific research may 
be listed in terms of: 

1) Size. — It is just large enough to be handled easily. Its organs 
are large enough to permit almost any kind of operation to be performed. 
It is small enough that large numbers, literally hundreds, may be housed 
in the space that would hold only a few dogs. 

2) Diet. — Its diet is almost the same as man's. For this reason much 
of our modern knowledge of nutrition has come from experiments on rats. 

3) Physiology. — Its physiology of nerves, muscles and glands is much 
the same as man's. It has a stable nervous system so that results obtained 
at one time may be repeated at another. This is not true of the rabbit. 

4) Reproduction. — It breeds very readily under conditions of domes- 
tication. Many litters can be obtained in a short time and at a small 

5) Resistance to infection. — It has a high degree of resistance to many 
kinds of infection which greatly adds to its value as an experimental 
animal, since time-consuming aseptic technique need not be used. 

6) Handling. — Since, after domestication, the Norway rat remains 
tame even with only occasional handling, workers can remove it from 

Vol. 15, No. 3, December 1954 

730 proceedings: SYMPOSIUM ON 25 YEARS op 

cages, and inject it, etc., without running the risk of having it escape. 
In this way it contrasts sharply with the Alexandrine rat which because 
of great fleetness and agility cannot be handled at all. 

My acquaintance with the wild Norway rat stems from work carried 
on during the second world war. Before that time I had scarcely seen 
more than a dozen or two wild rats in my entire life. The war work was 
concerned with poisons and baits for wild rats to be used in a quick city- 
wide extermination campaign, should such an operation be needed (5). 
For this work we had to capture wild rats and bring them to the laboratory 
for toxicity and poison acceptance tests and for physiological studies on 
the action of various poisons, involving investigation of the endocrine, 
nervous, and circulatory systems. It quickly became obvious that 
despite the external similarities between the wild and domesticated 
Norway rat, many internal differences existed. A systematic study was 
undertaken to define these differences and to determine what general 
biological significance, if any, they have. 

During the past 15 years many thousand wild Norway rats — rats 
captured from alleys, cellars, and farms — have been used in my laboratory 
for all kinds of investigations. Simple methods have been devised for 
trapping them alive and for holding them for injection and examination 
without the use of anesthetics. More than a thousand rats have been 
used for experiments on the effects of such operations as adrenalectomy, 
hypophysectomy, and pancreatectomy. An excellent opportunity was 
thus afforded for becoming well acquainted with this animal. The most 
important single instrument in use for handling these wild animals is a 
device designed by Dr. John T. Emlen for holding unanesthetized rats (6). 
It has made it possible to handle the fiercest, wildest rats without danger 
and almost as readily as tame, domesticated rats. A modified rabbit trap 
is used in trapping the wild rats (7). We have caught as many as 300 
rats from one square block in Baltimore over a 10-day period. The traps 
are inexpensive and are now being widely used for trapping wild rats 
in other cities. 

Comparisons were made between captured wild rats and domesticated 
rats from our colony, which has been in existence for over 30 years. Our 
original albino stock came from the Wistar colony and about 28 years ago 
a few piebald and hooded rats were added from the colony of Dr. E. V. 
McCollum. Since then no new strains have been added and from the 
beginning, the diet and external conditions have remained the same. 
In physiological responses as well as in weight of organs and glands, our 
rats are very much like those from other laboratory colonies. No con- 
scious effort has been made to breed the rats for auy special character- 

The wild rats were trapped in the city of Baltimore or on outlying 
farms. Some were bred in the laboratory and studies were made of their 

Part of the data presented here has been reported in detail in earlier 
papers and part is being reported for the first time. 

Journal of the National Cancer Institute 


Anatomical Differences 

Here I shall pick out only a few of the more striking differences. Some 
organs have become smaller during domestication. For instance, as was 
first shown by Watson (8) and Donaldson (9), and later by Rogers and 
Richter (10), the adrenals may be one-third to one-fifth as large as in the 
wild rat; a reduction of cortical tissue accounts for the entire difference. 
The preputial glands have likewise become much smaller (11) as have also 
the liver, spleen, brain, heart, kidneys and pancreas. Other glands have 
become larger, for instance the hypophysis and thymus, and other organs 
have developed earlier, such as the gonads (12). The number of Peyer's 
patches in the intestines of each rat has increased from 16.3 to 18.9 (IS) 
and the number of fungi-form papillae on the tongue of each rat has been 
markedly reduced from an average of 217.9 to 178.3 (H). The changes in 
thyroid weights are not definite. 

Physiological Differences 

Here again I shall select only a few outstanding differences. The 
domesticated rat has a lower metabolic rate as is shown by its lower food 
intake per kilogram of body weight and also by its lower water intake. 

It has a lower resistance to poisoning with the several compounds 
tested, especially thiourea (15). 

On a normal diet, domesticated rats show audiogenic fits when exposed 
to sounds of high frequency but it has not yet been possible to put a 
single wild rat into a fit in this way (16). On a magnesium-deficient diet, 
all domesticated rats show audiogenic fits and die within a few days; 
on the other hand, although magnesium-deficient wild rats show fits they 
do not die as a result (17). 

Behavioral Differences 

Apart from the differences in behavior described above, a few others 
may be mentioned. 

In the first place, domesticated rats have lost a very characteristic 
high-pitched squeak or squeal that is characteristic of the wild rat when 

Domesticated rats do not huddle together as wild rats do. Wild rats 
pack themselves very closely together in a corner against the walls of a 
cage and remain motionless literally for hours. 

With few exceptions, wild rats kill small rats or mice as rapidly as they 
are fed into their living cages. Some rats repeatedly bite almost any part 
of the bodies of such small rats, but others bite only once and then through 
the spinal cord at the base of the brain. Except under special circum- 
stances, domesticated rats pay little or no attention to small rats or mice 
that are thus introduced into their cages. One such special circumstance 
occurred when a domesticated lactating mother that had nursed a litter 
of 15 rats for 14 days, in one night, killed all of the babies — each one 

Vol. 15, No. 3, December 19S4 

732 proceedings: symposium on 25 years of 

with a single finely executed bite through the spinal cord at the base of 
the brain. 

In both strains of rats, fasting increases the spontaneous activity as 
measured in revolving drums, but the increase in activity, during fasting, 
is over 4 times as high in the wild as in the domesticated rats (18). It 
would thus appear that in a free environment the wild rats would have a 
much greater chance of coming into contact with food, through their 
greater activity. 

Studies in our so-called "fighting chamber' ' brought out a further 
difference. This chamber consists of a box 12 by 12 by 18 inches with a 
glass front and a floor made of parallel iron rods spaced at intervals of three 
quarters of an inch, and alternately wired to the opposite poles of an 
induction coil. A pair of wild rats placed in such a cage and given a single 
shock will start fighting at once, and only an occasional shock suffices to 
keep them fighting. Evidently each one holds the other responsible for 
the inflicted pain. However, a pair of domesticated rats will not fight, no 
matter how severe the punishment, but each one simply tries to escape. 

Wild rats are much more difficult to poison since they are so suspicious 
of all changes in their food. When repeatedly threatened with poisoning, 
they may ultimately refuse all food and starve themselves to death (19). 
We have never seen a domesticated rat starve itself to death in this way. 

These two strains of rats have a very different reaction to physical 
restraint: that is, being held in the experimenter's hand and thus prevented 
from biting, scratching, or escaping. For the wild rats, as for most other 
wild animals, this is obviously a terrifying situation. They struggle 
violently for a few minutes and then often seem to become limp. A few 
rats react so violently that they actually die within minutes. In marked 
contrast, domesticated rats react only mildly, if at all, to this restraint 
and tend to accept it with little resistance. 

Electrocardiographic records taken while the rats are being restrained 
show, in the case of wild rats, marked slowing of the heart — a definite 
bradycardia, and little or no change in the domesticated rats. Thus it 
appears that, as compared with wild rats, domesticated rats are much less 

That the wild rats react so much more violently to physical restraint 
and actually succumb in some instances, does not mean that they are less 
able to meet stress in general. It is merely that restraint bothers them 
much more than it does the domesticated rat. 

Pattern of Domestication 

Do the changes in anatomy, physiology and behavior seem to follow any 
definite pattern? We believe that a pattern is emerging from observations 
made so far, particularly on the adrenals, gonads or hypophysis. 

We have found that during the process of domestication the adrenals 
apparently are becoming less important to the rat, whereas the gonads are 
becoming more important; and likewise that the hypophysis may be 
becoming more important (12). 

Journal of the National Cancer Institute 


What evidence do we have for these statements? 

First of all, the adrenals are one-third to one-fifth as large in the domesti- 
cated as in the wild rats. Cortical tissue accounts for the entire difference. 
By lipid stains it was shown that in the domesticated rats the outer layer 
of the cortex, the glomerulosa or salt-regulating layer, is much less active; 
and that likewise the next layer, the fasciculata or corticoid-secreting 
layer, whose secretion helps the animal to meet stress, is much less active 

Second, as is known, the glomerulosa layer has salt regulatory functions, 
in that secretions from this layer help to conserve salt — preventing salt 
from being washed out of the body into the urine. After adrenalectomy 
both domesticated and wild rats voluntarily drink more salt solution. 
However, the domesticated rats survive and remain in good health while 
drinking only slightly increased amounts of salt solution, but the wild rats, 
on the contrary, do not survive adrenalectomy after treatment with any 
amount of salt. They need in addition large amounts of cortical hormone, 
and even then they do not survive with any consistency {21). 

That in normal wild rats, the adrenal functions far more effectively is 
shown by the observation that wild rats are able to survive long periods of 
time, even months, on a very low sodium diet, without any visible effects, 
whereas domesticated rats succumb in a short time. 

The wild rats conserve salt so well that salt, when not accompanied by 
adequate amounts of water, is actually toxic to them, even in very small 
amounts. Domesticated rats are able to take much larger amounts of salt 
with impunity. 

Many workers (22) have established that the changes in the amount of 
ascorbic acid in the adrenals give a measure of the severity of the reaction 
of an animal to various forms of stress. The greater the drop in ascorbic- 
acid levels, the more severe is the reaction to stress. 

Domesticated and wild rats were subjected, by Dr. James Woods, to the 
same form of stress: either exposure to cold, fighting, or exhaustion from 
swimming {23). In the wild rat none of these forms of stress had any 
detectable effect on the amount of ascorbic acid in the adrenals, whereas 
in the domesticated rat it either reduced the ascorbic-acid content to a 
very low level or else eliminated it altogether. In wild rats, the ascorbic- 
acid content could be depressed only by large doses of ACTH. These 
results would thus indicate that the wild rat is better able to withstand 
various forms of stress — because of the more active adrenals. 

This evidence indicates that the adrenal secretions necessary both for 
salt metabolism and for reaction to stress are more effective in the wild 
than in the domesticated rat. 

It has already been reported that the gonads develop earlier in the 
domesticated than in the wild rat, and that domesticated rats mate and 
reproduce more freely and more often. In the domesticated rat gross 
bodily activity, as measured in revolving drums, has been shown to be 
dependent on gonadal secretion, since after gonadectomy the rats become 
very inactive (24). The average daily activity may drop from 18,000 

Vol, 15, No. 3, December 1954 

734 proceedings: symposium on 25 years op 

revolutions to only a few hundred. However, in the wild rat, gonadectomy 
has no detectable effect on the level of running activity, since the rats are 
quite as active after as before surgery {12). Apparently secretions from 
the larger adrenals keep the wild rats active. That this is so, was shown by 
the results of experiments in which domesticated rats were started on 
cortisone immediately after gonadectomy. They remained active after 
gonadectomy just as the wild rats do without treatment. 

Observations have been made by Davis {25), and Davis and Emlen {26), 
and others on wild rats caught in the field, on the time of opening of the 
vagina, incidence of pregnancy, and the time of the signs of first pregnancy. 
Comparison shows that the domesticated rats mature at an earlier age and 
show a higher degree of fertility. All of this evidence thus indicates that 
gonadal secretions have become more important with domestication. 

We have seen that the hypophysis has become larger in the domesticated 
rat. David Wood has found that hypophysectomy, which makes a domes- 
ticated rat almost totally inactive, has a much less depressive effect on 
some wild rats; whereas others, especially very old and heavy rats, do not 
survive the operation {27). 

It is possible that in the domesticated rat the hypophysis has become 
enlarged in an effort to correct for the failing secretion of the adrenal 
glands. Thus we may see here a pattern of less active adrenals, more active 
gonads, and a more active hypophysis. Further studies are in progress on 
that relationship by David Wood. 

At this point I should like to say a word about the neurology of domes- 
tication. What changes, if any, have occurred in the nervous system? 
A few observations on the effects of brain lesions in wild and domesticated 
rats have been made. Removal of the frontal poles of the brain makes 
domesticated rats just as savage as wild rats {28). This is best shown by 
their reaction when confronted with mice or small rats. One such domes- 
tic rat killed small rats just as fast as they could be fed to her and always 
with the same clear-cut bite — through the brain stem at the base of the 
head, with much the same technique used by wild rats. These results 
would indicate that during domestication a change may have occurred in 
this part of the brain. Secondly, it has been shown by J. Woods that 
wild rats can be made just as tame as domesticated rats by lesions placed 
in the amygdaloid area of the brain {29) . We are now studying domesti- 
cated rats with frontal lobectomies and wild rats with amygdaloid lesions 
to determine whether these operations are followed by changes in the 
endocrine glands and other organs, as well. 

Domestication Mechanisms 

How can we explain these various changes that have occurred during 
the process of domestication of the rat? 

The wild Norway rat lives in an environment in which it must con- 
stantly be on the alert and ready to fight for its existence. It has to 

Journal of the National Cancer Institute 


defend itself against all kinds of enemies: other rats, dogs, cats, owls, 
and snakes, as well as man. It is a fierce, aggressive, suspicious animal 
that attacks at the least provocation and in captivity, takes advantage 
of the least opportunity to escape, remains suspicious and tense, and breeds 

In marked contrast, the healthy domesticated Norway rat is tame, 
gentle, and trusting, does not bite unless frightened or hurt, and makes no 
attempt to escape. It lives placidly in the controlled environment of the 
laboratory where food, water, shelter and safety are constantly assured. 
Its only contributions to its own survival are its feeding, drinking, groom- 
ing and mating activities. It reproduces at an early age and with a rapid 
rate. Like other domesticated animals it has shown numerous mutations. 
Castle, a pioneer worker in this field, listed 23 strains that breed true (30) . 

We believe that selection has played the most important part in the 
process of domestication. By selection we mean here not the natural 
selection of wild rats in their natural habitat, where the fiercest, wildest 
and strongest — the fittest for that type of environment — survive, but 
selection in the artificial environment of the laboratory where the fittest 
for this type of environment survive: those that are most gentle and fertile. 
The two life stages in which the selection process has the most effect are 
during mating and nursing. In captivity wild rats do not mate well. In 
only a few out of many instances when rats are put together does a preg- 
nancy result. It is likely that only the tamest of the rats mate. This 
is probably true of each succeeding generation as well — so that the tamer 
rats are more apt to propagate their own characteristics, including pre- 
sumably the smaller adrenals, etc. After the young are born, wild 
mothers, with rare exception, eat their entire litter, so that at this stage 
a severe screening process occurs. The rats that do survive these two 
stages must be the least apprehensive, and so through successive genera- 
tions they should produce progressively tamer animals. 

How long did it take to domesticate the rat? This we do not know. 
After 25 generations the wild rats of King and Donaldson had not yet 
quite reached the various weight levels characteristic of domesticated 
rats (31, 32). 

So far our studies have been concerned almost entirely with differences 
between the captured wild rat and the tame domesticated rat. Little 
or nothing has been done with the underlying genetic problem to establish 
just what part of the difference depends on experience, and what part on 
inheritance. To fill this gap 1) we have placed domesticated rats in a 
wild free environment and are observing the effects on behavior, anatomy 
etc., and 2) we are planning to study hybrids of the two strains. Results 
of preliminary observations indicate that the hybrid offspring changes to 
become more like the domesticated than the wild strain, and 3) we are 
planning also to study rats born from fertilized eggs of one strain placed 
in the uteri of the other strain. This should clarify the role of environ- 
mental influences. 

Vol. IS, No. 3, December 1954 

736 proceedings: symposium on 25 years op 

Domestication in the Rat and Civilization in Man 

Finally I want to make a few remarks about what light, if any, these 
observations throw on the domestication of man. I trust Dr. Snyder 
will forgive this transgression into his part of the program. 

We know that man, like the wild Norway rat, originally lived in an 
environment in which he had to search for his food, provide his own shelter, 
and fight for his mates — an environment, in short, in which his fitness, 
hence his survival, was measured by his physical activity, aggressiveness 
and ability to withstand violent changes. But with the growth of com- 
munities and the consequent increase in security a new environment 
developed in which man was protected and a livelihood was almost guaran- 
teed. Man must thus have worked out a controlled environment for 
himself in which a transformation occurred, somewhat like that under- 
gone by the Norway rat in its adaptation to colony life in the laboratory, 
resulting in an increase and perhaps even the predominance of the so- 
called weaker, or the milder, better adjusted individuals. 

Here, just as in the domestication of the rat, a selective process — the 
selection of the fittest for this type of environment — must have played 
the most important part. This process has gone far in our present-day 

It has not been possible to obtain gland weights of modern and prima- 
tive man — especially of the adrenals and gonads to determine what changes 
if any have occurred. There are, however, indications that marked 
changes may have occurred in these glands. At the present time it has 
been estimated that over 12 million individuals in this country suffer 
from hypertension, rheumatism, arthritis, etc., all of which may have 
their origin in deficient functions of the adrenal glands since they respond 
so remarkably to treatment with cortisone and ACTH. That a perma- 
nent deficiency underlies most of these diseases is known from the fact 
that treatment with cortisone and ACTH gives relief as long as it is 

Similarly, we know that many individuals suffer from neoplastic dis- 
eases, some of which apparently have their origin in or are greatly influenced 
by hyperactivity of the gonads, since reduction or elimination of the 
gonadal secretion through gonadectomy so remarkably arrests the growth 
of these neoplasms, and treatment with sex hormones greatly accelerates 
the growth of this group. Could this be an indication of a shift in the 
direction of domestication? 

Likewise, modern man is eating more and more salt, and in fact, the 
eating of large amounts of salt has almost become a characteristic of civil- 
ization. The domesticated rat likewise eats and is able to tolerate very 
large amounts of salt. Are the individuals who do not tolerate salt those 
who have not been so much affected by the domestication process, and 
are more like the wild rat? 

Thus, in summary, the Norway rat, the first animal to be domesticated 
for experimental purposes, has during the course of its domestication and 
during its transition from the free environment of its wild habitat to the 

Journal of the National Cancer Institute 


controlled conditions of the laboratory, undergone marked anatomical, 
physiological, and behavioral changes. This study has offered for the 
first time, the unique opportunity of observing a shift of the interrelations 
of glands rather than of the function of individual organs. The secretion 
from one gland, the adrenal, has become less important and the secretion 
from the gonads more important. The secretions of the hypophysis have 
become more important but their relation to the secretions from the other 
two glands has not yet been definitely defined. It is possible that this 
general shift will be found to involve other glands as well, and also that 
the various forms of changes in behavior will eventually fall into relation- 
ship with this general pattern of domestication changes. The possibility 
must thus be considered that similar changes and shifts may have occurred 
in various degrees in man during the transition from his original free envi- 
ronment to the highly protected and controlled environment of modern 


(1) Donaldson, H. H.: The Rat. Memoirs of the Wistar Institute, No. 6, 2d 

ed., 1924. 

(2) Philipeaux, J. M.: Note sur l'extirpation des capsules surrSnales chez les rats 

albinos (Mus rattus). Compt. rend. Acad. sc. 43: 904-906, 1856. 

(3) Conklin, E. G.: Biographical Memoirs of H. H. Donaldson (1857-1938). Nat. 

Acad. Sc, 1939. 

(4) Hatai, S. : On the appearance of albino mutants in litters of the common Norway- 

rat, Mus norvegicus. Science 35: 875-876, 1912. 

(5) Richter, C. P.: The development and use of alpha-naphthyl thiourea (Antu) 

as a rat poison. J. A. M. A. 129: 927-931, 1945. 

(6) Emlen, J. T., Jr.: Device for holding live wild rats. J. Wildlife Mgt. 8: 264- 

265, 1944. 

(7) Richter, C. P., and Emlen, J. T., Jr.: A modified rabbit box trap for use in 

catching live wild rats for laboratory and field studies. Pub. Health Rep. 60: 
1303-1308, 1945. 

(8) Watson, C: A note on the adrenal gland in the rat. J. Physiol. 35: 230-232, 


(9) Donaldson, J. C: Adrenal gland in wild gray and albino rat: Cortico-medullary 

relations. Proc. Soc. Exper. Biol. & Med. 25: 300-301, 1928. 

(10) Rogers, P. V., and Richter, C. P.: Anatomical comparison between the ad- 

renal glands of wild Norway, wild Alexandrine and domestic Norway rats. 
Endocrinology 42: 46-55, 1948. 

(11) Hall, C. E.: Comparison of the preputial glands in the Alexandrine, the wild 

and the domestic Norway rat. Proc. Soc. Exper. Biol. & Med. 69: 233-237, 

(12) Richter, C. P., and Uhlenhuth, E. H.: Comparison of the effects of gonad- 

ectomy on spontaneous activity of wild and domesticated Norway rats. En- 
docrinology 54: 311-322, 1954. 

(13) Richter, C. P., and Hall, C. E. : Comparison of intestinal lengths and Peyer's 

patches in wild and domestic Norway and wild Alexandrine rats. Proc. Soc. 
Exper. Biol. & Med. 66: 561-566, 1947. 

(14) Fish, H. S., and Richter, C. P.: Comparative number of fungiform and foliate 

papillae on tongues of domestic and wild Norway rats. Proc. Soc. Exper. 
Biol. & Med. 63: 352-355, 1946. 

(15) Dieke, S. H., and Richter, C. P.: Acute toxicity of thiourea to rats in relation 

to age, diet, strain and species variation. J. Pharmacol. & Exper. Therap. 83: 
195-202, 1945. 

Vol. 15, No. 3, December 1954 
316263—54 32 


(16) Griffiths, W. J., Jr.: Absence of audiogenic seizures in wild Norway and Alex- 

andrine rats. Science 99: 62-63, 1944. 

(17) : Audiogenic fits produced by magnesium deficiency in tame domestic 

Norway rats and in wild Norway and Alexandrine rats. Am. J. Physiol. 149: 
135-141, 1947. 

(18) Richter, C. P., and Rice, K. K.: Comparison of the effects produced by fasting 

on gross bodily activity of wild and domesticated Norway rats. In press. 

(19) Richter, C. P.: Experimentally produced behavior reactions to food poisoning 

in wild and domesticated rats. Ann. New York Acad. Sc. 56: 225-239, 1953. 

(20) Mosier, H. D., Jr.: Histological study of the adrenals of wild and domesticated 

Norway rats. In preparation. 

(21) Richter, C. P., Rogers, P. V., and Hall, C. E.: Failure of salt replacement 

therapy in adrenalectomized recently captured wild Norway rats. Endocri- 
nology 46: 233-242, 1950. 

(22) Sayers, G., and Sayers, M. A.: The pituitary-adrenal system. Ann. New York 

Acad. Sc. 50: 522-529, 1949. 

(23) Woods, J. W.: Some observations on adrenal cortical functions in wild and dom- 

esticated Norway rats. Doctoral Thesis, Johns Hopkins University, 1954. 

(24) Wang, G. H.: The relation between "spontaneous" activity and oestrous cycles 

in the white rat. Comp. Psychol. Mon. 2: 1-27, 1923. 

(25) Davis, D. E.: The weight of wild brown rats at sexual maturity. J. Mamm. 30: 

125-130, 1949. 

(26) Davis, D. E., and Emlen, J. T., Jr.: The placental scar as a measure of fertility 

in rats. J. Wildlife Mgt. 12: 162-167, 1948. 

(27) Wood, D. E.: Behavioral and metabolic studies on the hypophysectomized wild 

and domesticat'ed Norway rat. In preparation. 

(28) Richter, C. P., and Hawkes, C. D.: Increased spontaneous activity and food 

intake produced in rats by removal of the frontal poles of the brain. J. Neurol. 
& Psychiat. 2: 231-242, 1939. 

(29) Woods, J. W.: Taming effect of amygdaloid lesions in wild Norway rats. In 


(30) Castle, W. E.: The domestication of the rat. Proc. Nat. Acad. Sc. 33: 109- 

117, 1947. 

(31) King, H. D.: Life processes in gray Norway rats during fourteen years of cap- 

tivity. Am. Anat. Memoirs No. 17, 1939. 

(32) King, H. D., and Donaldson, H. H.: Life processes and size of the body and 

organs of gray Norway rat during ten generations in captivity. Am. Anat. 
Memoirs No. 14, 1929. 

The Effects of Selection and Domesti- 
cation Upon the Behavior of the Dog u 2 

J. P. Scott, 3 Division of Behavior Studies, Roscoe 
B. Jackson Memorial Laboratory, Bar Harbor, 

Richter has shown here and elsewhere (1) how laboratory domestication 
has produced drastic changes in the behavior of the Norway rat. These 
changes, which have their counterparts in almost all domestic animals, 
appeared within a few generations, and the rat is still less than a century 
old as a domestic species. It is therefore of considerable interest to com- 
pare it with the dog, which has probably been domesticated for something 
like 80 centuries. Data on the rat give us information regarding the ease 
with which the changes of domestication can take place, and data on the 
dog give indications as to whether or not the process is a progressive one. 

Most of the typical changes of domestication are seen in an exaggerated 
form in the dog. The most obvious effect is increased variability com- 
pared to the wild species, both in anatomical characteristics and in 
behavior. Some breeds of dogs like the great Danes and St. Bernards 
are larger than wolves; others like Chihuahuas and Pekingese are much 
smaller. Some breeds like the greyhounds can run faster than wolves and 
others like dachshunds are obviously much slower. Most breeds of dogs 
are less aggressive than wolves but others are much more so; e.g., certain 
terriers will attack bears and mountain lions, while a wolf pack will only 
attempt to drive a bear away if it approaches the den. 

A second effect which is seen in almost all domestic animals is a tendency 
toward early sexual maturity and increased fertility. Wolves do not 
become sexually mature until the second year, whereas the vast majority 
of dogs mature before the age of one year. The average litter size in 
wolves is about 7 and the largest recorded litter is 14 [Young and Gold- 
man (#)]. The average litter size for all dogs is probably about 6, but 
the average in a breed such as the great Dane runs 9.1 and a litter of 16 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 29, 1954. 

* This paper is based on a research project in genetics and social behavior begun in 1945. 

3 Thanks are due to the many individuals who have contributed to the paper, particularly, Dr. C. C. Little, 
under whose general direction the work was done and without whose knowledge and experience of dog breeds, 
progress would have been slow and difficult, and Dr. John L. Fuller, who bears co-responsibility for the project. 
The author ajso wishes to express his appreciation to the many research assistants who have helped gather the data, 
including Edna DuBuis, Mary-'Vesta Marston, Margaret Charles, and others (especially Albert Pawlowski and 
Ann Causey, who assisted greatly in the preparation of this paper). Thanks are also due to Dr. Eloise Gerry who 
contributed our original stock of basenji dogs. 


Journal of the National Cancer Institute, Vol. IS, No. 3, December 1954 

740 proceedings: symposium on 25 years of 

was recorded in one group of 48 [Little aud DuBuis (3)]. In many smaller 
breeds of dogs the average litter size is lower than that in wolves, and 
increased fertility is therefore typical only of certain breeds. 

A third effect is a modification of the process of socialization with 
respect to human beings so that it is easier to get domestic animals to form 
attachments to people. This may also be looked at as a tendency to 
reduce the flight distance, and Richter (1) has demonstrated a remarkable 
change in the adrenal glands which accompanies this effect in the rat. In 
the dog there is also a tendency to develop individuals that are positively 
responsive to human beings, and examples of both of these effects will be 
treated in detail below. 

Most widely used domestic animals belong to species which have a 
considerable range of adaptability and this may be one reason why certain 
species rather than others have been domesticated. However, there is a 
tendency under domestication to select individuals for particular kinds 
of behavioral adjustment and thus limit or extend the powers of adaptation. 
This process appears to have been carried further in the dog than any 
other species, while only limited experiments have been done with rats. 

The -Ancestry of Dogs 

The family Canidae. — To this family belong the foxes (Vulpes, Urocyon, 
Alopex); the so-called wild dogs including Lycaon y the African hunting 
dog; Icticyon, the South American bush dog; Cuon, the dhole of India; 
and, finally, the true dogs and wolves belonging to genus Cams. Accord- 
ing to Matthew (4) the foxes and wolves have a common ancestor in the 
Miocene, but the relationship with the wild dogs is much more remote 
and a common ancestor is only found in Cynodictus of the Oligocene. 
From the geological evidence it is apparent that the so-called wild hunting 
dogs are less closely related to the domestic dog than are the foxes. 

The genus Canis. — This genus includes three groups of animals: the true 
dogs, Canis familiaris and Canis dingo; the wolves such as the northern 
wolf, Canis lupus and the prairie wolf or coyote Canis latrans; and the 
jackals such as Canis aureus. Matthew (4) traces dogs and wolves back 
to a common ancestor in Pleistocene times when wolves, coyotes, jackals 
and foxes are found essentially in their modern forms. Allen (5) postu- 
lated a small species of wolf which is now extinct as the direct ancestor of 
the dog, but, so far, no such remains have been discovered. All the 
anatomical evidence indicates that the dog is a domesticated variety of 
wolf and that the closest related living species is probably Canis lupus, 
which is found in the arctic and temperate regions of both Eurasia and 
North America. 

At various times authors have postulated other species such as Canis 
latrans, the coyote, and Canis aureus, the jackal, as ancestors either of 
dogs or certain breeds of dogs, but the evidence seems to be based chiefly 
on superficial resemblances in size. Detailed examinations of skulls and 
teeth do not confirm these ideas. 

Journal of the National Cancer Institute 


Archeological evidence. — For the most part archeologists have been 
interested chiefly in human history and very few studies have been made 
of dogs except as the incidental accompaniment of human remains. In 
Europe the most extensive recent investigation has been made by Degerbol 
(6) who studied the prehistoric dogs of Denmark. The older remains in 
Denmark are found in the last part of the Old Stone Age in a transition to 
the Neolithic Age but may possibly be older as indicated by pollen analysis. 
This discovery has been dated as possibly 6,000 to 8,000 years B. C. and 
at the present time can be only accurately placed with regard to Azilian 
culture. The fragments of dogs which were found belong to two types, 
and Degerbol finds that the larger one is quite similar in measurements to 
Eskimo sled dogs and also to remains found in later European cultures 
which have been called Canis Jamiliaris inostranzewi. The smaller form 
he finds closely similar to a dog whose bones were discovered in Lake 
Ladoga in Russia and also associated with the Swiss lake dwellers and 
which has been named Canis Jamiliaris palustris. Allen (5) considers that 
both these forms are closely similar to the modern Eskimo dogs. These 
earliest remains differ from wolves chiefly in the fact that the dogs have, 
like modern breeds, smaller-sized teeth. Haag (7) points out that dogs 
were buried as early as 6,000 B. C. in Egyptian tombs and that these re- 
mains may be as early as the Denmark finds. It is to be hoped that future 
studies with the new techniques of radioactive carbons and pollen analysis 
will produce a clearer picture, but it is evident that the dogs were domesti- 
cated on the continent of Europe at a very early age. 

In North America Allen (5) found historical evidence that when the 
white man arrived in North America dogs were distributed among the 
native Indian tribes from Alaska to Tierra del Fuego and that there was 
a great deal of variability in the different regions. There was the Eskimo 
dog, the common Indian dog of moderate size, which was found in at 
least eight different varieties or breeds, and the small Indian dogs which 
included at least five different varieties. Of these only two have survived ; 
the Eskimo dog, which is very similar to the Eskimo dog of Europe, and 
the Mexican hairless dog. Haag (7) has recently studied skeletal re- 
mains of dogs from various regions of North America and, using modern 
methods of statistical analysis, comes to the conclusion that at least eight 
different populations can be separated. Those close together geograph- 
ically, as in Kentucky and Alabama, tend to be similar. It may be 
concluded that each Indian tribe possessed dogs, sometimes of more than 
one variety, and that the dogs tended to vary from tribe to tribe. 

By contrast, only one form of dog was found in Australia, the dingo. 
Canis dingo apparently existed both in wild and semidomestic forms and 
probably was brought by the aborigines to Australia where it apparently 
became a separate species by isolation. 

The dogs of Africa have not been subjected to extensive scientific study 
but were found in various breeds all over the continent by recent white 
explorers. In many places, European breeds have since been imported 
and mixed with native populations. One native breed, the African 

Vol. 15, No. 3, December 1954 

742 proceedings: symposium on 25 years of 

barkless (basenji) was recently imported into England without admixture, 
and has been used in extensive studies to be reported here. 

All of the evidence indicates that the dog was domesticated probably 
from Canis lupus or a very similar species of wolf in Northern Europe or 
Asia about 6,000 to 8,000 years B. C. Once domesticated the use of dogs 
spread very rapidly over the world and in modern times there appears to 
have been no humanly inhabited continent from which they were absent. 
Each primitive tribe had its own population of dogs the type of which 
differed from tribe to tribe. Such conditions provide the small isolated 
breeding populations postulated by Wright and others for rapid evolution. 
This, combined with the undoubted existence of differential selection, will 
account for the remarkable diversity in modern dog breeds. In historical 
times there has been a great deal of transporting of dogs from one area 
to another by explorers, immigrants and dog fanciers with the resulting 
possibility of mixtures of the populations. On the other hand, in civilized 
countries, there has been an effort to maintain separate breeds and over 
100 of these are recognized by the American Kennel Club. 

The Domestication of Wolves 

The social life of wolves. — We cannot, of course, discover what actually 
happened when the wolf was first domesticated but we can deduce certain 
things from two sources: the study of the behavior of wild wolves and 
the results of attempting to tame them. An excellent study of wolf 
behavior has been done by Murie (8), and Young and Goldman (2) have 
collected observations made by trappers and hunters. A detailed com- 
parison of these descriptions with various domestic breeds of dogs shows 
that wolves and dogs exhibit the same basic traits of behavior [Scott (9)]. 
There is a great deal more variability in dogs, but only one new trait 
seems to distinguish dogs universally from their wild relatives. This is 
the tail carriage, which in dogs varies from a sickle shape to a tight curl 
but in wolves is almost straight when relaxed. Other differences be- 
tween dogs and wolves have been alleged such as the fact that wolves 
have no bark, but all careful observers agree that they do bark, although 
not as much as some breeds of dogs. 

The wolf puppies have a relatively long period of dependency. They 
are weaned about 7 or 8 weeks of age, but continue to receive food from 
the adults for a long period and first begin to hunt for themselves at about 
4 months of age. The wolf puppies are left at the den while the parents 
and other members of the pack go out to hunt and large quantities of 
food are brought back. Some of this is vomited for the young and, in 
addition, large pieces of meat and bones may be brought back and cached 
near the den. Young and Goldman (#) recount an incident where some 
150 pounds of disgorged beef were found near a wolf den. The wolves 
are scavengers as well as hunters and where they are allowed to live near 
human habitation are frequently seen around garbage dumps. 

The social group of the wolf is the pack, which seems to be based on a 
litter of animals which grows up together. Occasionally the pack may be 

Journal of the National Cancer Institute 


founded by a single pair and a litter which stays on as adults, but the 
young litters ordinarily move out of the territory to form new packs. As 
described by Murie, the home life of wolves is highly peaceful and coopera- 
tive. A certain amount of dominance is exhibited while feeding, but the 
members of the pack get along well together even when more than 
one male is present. Strange wolves, however, are violently repulsed 
and driven off. Various persons have described attempts to rear wolf 
cubs taken at a very early age and Murie found that such a female animal 
became a very tractable pet and that there was little difficulty in 
handling it, although caution had to be used with strangers. 

On these bases we may reconstruct the domestication of the wolf as 
follows: a primitive hunting tribe in Europe or Asia may very easily 
have fallen into a commensal relationship with wolves with the latter 
frequenting the village refuse heap and the human inhabitants finding it 
profitable to rob the wolf dens at times when food was scarce. At some 
point, a young wolf puppy was caught and adopted by the people and 
nursed and fed by them. This animal would consider the human group 
as its society and would be peaceful and tolerant toward them and as it 
grew older would have a tendency to go out and hunt and bring back 
food to the village. It would reject the wild wolves except at the time of 
mating and in due course other wolf puppies could be raised in the village 
and be socialized with respect to man. There would be, of course, the 
difficulty of telling the tame and wild wolves apart but, when the tail 
mutation occurred in the domestic form, an easy method of identification 
would be established and preserved. There would also be a tendency to 
select for certain other traits, particularly tameness and fertility. Once 
the domestic strain was established, the use of these animals would tend 
to spread rapidly to other tribes. 

The Effects of Domestication Upon the Behavior of the Dog 

Breeds studied. — In order to study the effects of heredity in causing 
the variable behavior seen in dogs, 5 different breeds were selected for 
special study, and representative individuals were raised in a similar 
environment which included a regular program of training and testing 
from birth up to the age of 1 year. The requirements were that these 
breeds include as wide a variety of behavioral types as possible, that they 
be of medium size so that they would be easy to handle, and that they 
would be reasonably hardy and fertile so that reliable genetic results 
could be obtained. It was felt that physical characteristics such as short- 
leggedness produced an obvious effect upon behavior which was unneces- 
sary to demonstrate, and the breeds selected were reasonably normal 
with regard to sensory and motor development. 

Dog fanciers classify the dog breeds according to the use to which they 
are put and this functional classification tends to result in grouping some 
breeds which are genetically unrelated, particularly if dogs of widely 
different geographical origin are included. The final selection included 
4 breeds which originated in the British Isles (cocker spaniels, beagles, 

Vol. 15, No. 3, December 1954 

744 proceedings: symposium on 25 tears of 

Shetland sheep dogs, and wire-haired fox terriers, which consequently 
have in the past had opportunities for interbreeding, and one breed from 
Central Africa, which according to all records should have been completely 
separated for many centuries. The African barkless breed or basenji, was 
selected for crossbreeding with the cocker spaniel. These two breeds and 
the tasks which they ordinarily perform may be briefly described as follows. 
The African basenji originally came from the Belgian Congo and our 
animals are all descended from 5 individuals imported into England in 
1935-1940 [Williams (10)]. Little information is available about their 
use in the native villages. It is reported that they are used as hunting 
dogs chiefly in the pursuit of game, but also are able to trail. They are 
reported to be related to the Egyptian greyhound and are physically 
somewhat similar. It is probable also that their ancestors ran loose in the 
villages and acted as scavengers. They are long-legged, muscular, and 
have considerable skill at climbing and manipulating objects with their 
paws. They are specialized in many respects, but appear to be primitive 
in that there is an annual, seasonal breeding cycle which, however, takes 
place in the autumn instead of in the spring as with the wolf. Basenjis 
bark very seldom but, when frustrated, make a variety of howling noises 
which are peculiar to the breed. As far as known, they have never been 
crossed with European breeds. 

The cocker spaniel is mentioned in early English historical records. 
They are related to the whole group of bird dogs and other spaniels, and 
particularly to the setters which were originally used for setting birds 
for the net and were taught to crouch down when the birds were located, 
in order to be out of the way of the net. They have also been selected 
as retrievers and are reported to have been used in connection with the 
medieval sport of falconry in which they found the birds and flushed them 
so that the hawks could attack. They are very tolerant in groups and 
in modern times have been very successful as house pets as well as bird 

The plan of genetic analysis. — At this point it will be well to introduce 
a description of the experimental methods used. As stated before, the 
animals were raised in a standard environment in which most of the varia- 
bility should have reflected hereditary differences and which has been 
described in detail elsewhere [Scott and Fuller (11)]. It was found that 
all of the breeds showed a great deal of variability, a large part of which 
appeared to be hereditary since offspring of different matings gave dif- 
ferent results. In order to reduce this variability somewhat, the animals 
chosen from the parent strains for the crossbreeding experiment were 
descended from one brother X sister mating in the basenjis and from two 
matings of a single male with his sister and mother in the case of the cocker 
spaniels. No selection of these individuals was used except that the 
original pairs were vigorous and healthy animals. As it turned out later 
these did not necessarily illustrate the extremes of either breed in all 

Journal of the National Cancer Institute 


Reciprocal crosses were made between these two groups of siblings and 
an effort was made to obtain offspring from at least 4 different pairs in 
each case, giving two F x populations. F x males were backcrossed to the 
mothers so that backcross and Fi animals raised by the same mother 
could be paired. Finally, F 2 populations are being obtained from both 
crosses. It was planned to obtain at least two litters from each mating 
so that comparisons could be made between littermate and nonlittermate 
siblings. Because of the fertility and relatively long life of the dog, it 
has been possible to get relatively large numbers of animals from particular 

There was no way of predicting the experimental results in advance 
but the experiment was set up in such a way that the results could be 
subjected to a variety of statistical analyses. The cross was made in the 
form of a typical Mendelian experiment and can be analyzed in the usual 
way. The results can also be analyzed in terms of analysis of variance, 
correlation between relatives, and factorial analysis. Such incidental 
factors as litter size, age of mother and season of the year can also be 
analyzed. The only limitation of the experiment is the matter of num- 
bers, as it is impossible to raise and subject to detailed behavioral analysis 
more than about 12 litters, or 50 or 60 dogs per year. In many cases this 
results in the use of relatively small samples compared to many genetic 
experiments. We now have reasonably large samples in the F x and back- 
cross generations and in some of the early experiments have accumulated 
fairly large numbers of F 2 's. In the following paragraphs a sample of the 
genetic results so far obtained will be presented. 

Theoretical considerations. — There is every theoretical reason for an- 
ticipating that genetic differences in behavior should be affected by 
multiple factors rather than simple inheritance. A method for analysis 
of multiple factor data was devised by Castle and Wright (12) and Wright's 
formula for the analysis of backcross data is given below. The results of 


its use have recently been reviewed by Wright (18), who has pointed out 
the many difficulties that underlie its application. The original formula 
was based on the genetic assumptions that the parent stocks were homo- 
zygous and that there was no dominance. Heterozygosity does not 
affect the formula if there is no dominance but it does affect it if dominance 
is present. The modified formula for the backcross to the recessive, 
assuming dominance but no heterozygosis, is given below. 


M<Tbx 2 — o-jv) 

Vol. 15, No. 3, December 1954 

746 proceedings: symposium on 25 years of 

The basic mathematical assumption which lies behind these formulae 
is that the genetic factors are additive in theft effects and, as Wright (14) 
has pointed out, it is often necessary to transform the scale of measure- 
ments in order to bring out the additive effect. The formula also assumes 
that the effects of random environmental factors are additive. As 
pointed out in another paper [Scott (15)], this is not necessarily correct 
in the case of environmental factors acting upon behavior. In fact, there 
is every reason for believing the opposite to be true under many situations 
and, unless the environmental effects are additive, it is impossible to get 
a true estimate of genetic effects by subtracting the variance of one 
population from that of another. One solution for this difficulty is to 
find a situation in which environmental effects do seem to be additive. 
If there is a threshold of stimulation the fundamental physiological 
principle of summation of stimuli should hold and additive effects be 
produced at this point in the scale. 

There should be two general types of genetic situations. In one the 
threshold may be reached by only one extreme genetic class. In the 
other situation the threshold may be located somewhere near the center 
of the genetic distribution with several genetic classes on either side. 
Theoretical results have been calculated for both possibilities and the 
following method of analysis developed. 

a) Arbitrarily cut off the various experimental populations at two 
points: one which cuts off a terminal group and the other which most 
exactly separates the Fi into halves. 

b) For each population calculate the proportion of individuals found 
on one side of the dividing line. 

c) Subtract the proportion found in one parental population from that 
in the other. This gives a figure which may be represented as a distance 
and is a scale based on only one point in the original scale of measurement. 
The method requires only that environmental effects be additive at this 
point, which presumably represents a threshold. This distance also 
represents the genetic difference between the two parental stocks, and 
environmental variability has been eliminated. 

d) Find the position of the Fi and backcross generations relative to 
one of the parent stocks by subtracting the proportion found in it from 
the corresponding figures and dividing these figures by the distance 
between the parent stocks. This will give the relative position of the 
different populations on the scale referred to in c) . 

When the theoretical position of the Fi and backcross generations are 
calculated it is found that they are affected both by heterozygosis in the 
parent stocks and by the number of genetic factors, and graphs of these 
two effects have been prepared in order to show the effects of heterozy- 

However, one very interesting and useful relationship appeared from 
this analysis. The relative distance between the two backcross genera- 
tions is not affected by heterozygosis in a one-factor cross, even where 
dominance is involved. Therefore this figure, which is described by the 

Journal of the National Cancer Institute 



following formula, can be used as a simple test for a one-factor ratio. 

Bx\ — Bx2 

It is affected by heterozygosis when more than one factor is present, but has 
other interesting properties. If the threshold is reached only by the 
terminal class, and there is no heterozygosis, the distance between the 
backcrosses decreases regularly with the number of factors involved 
(table 1). On the other hand, if the threshold is located in the center of 
the genetic distribution, the distance between backcrosses increases 
regularly with the number of factors. The effect of additional factors 
is much greater when small numbers of factors are involved and ac- 
curate estimates would be difficult to make with more than four factors. 
It will be noted that when heterozygosis is present the estimate of the 
number of factors increases (text-fig. 1), so that an assumption of no 
heterozygosis will result in a minimum estimate of genetic factors involved. 

Table 1.- 

■Relative distance between 2 backcrosses when the data are arbitrarily divided 
into 2 parts, and populations are homozygous 



Threshold at 



Threshold mid- 
way between 
terminal classes 



. 125 



The above theoretical considerations assume that only one trait is 
being measured. It is very easy, in measuring behavior, to set up a unit 
which accidentally measures a combination of several traits, and if these 
are affected by different genetic mechanisms the results can be confusing. 
For example, an attempt was made to measure the fear responses of young 
puppies as indicated by avoidance, vocalization, postural and tail re- 
sponses, and all of these were added together for a single score of timidity. 
On detailed analysis it was found that the avoidance and vocalization 
scores appeared to be alternate expressions of fearful responses and could 
be added together. The postural responses, however, were not directly 
related. An animal might lie flat as a timidity response and he might also 
do so while getting ready to chew on the shoes and clothing of the observer. 
Consequently, the best genetic analysis is obtained when the data are 
divided up and analyzed in as small and discreet segments as possible. 
The following examples give the results of several such analyses. 

Avoidance and Vocal Reactions to Human Handlers at 5 Weeks 

This test is a measure of the process of socialization of young puppies 
toward their human handlers. As can be seen in table 2, the cocker 

Vol. 15, No. 3, December 1954 








I .3 





O .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 



Text-figure 1. — The distance between backcrosses relative to the distance between 
parent strains, in the case where a threshold cuts off a terminal class, and one of 
the parent strains shows heterozygosis. Roman numerals indicate the number of 
genetic factors involved. Note that the figure for 1 factor is unaffected by hetero- 
zygosis, and that with increasing heterozygosis an increased number of factors is 
required to account for the difference between backcrosses. 

spaniel and basenji breed show wide differences in this respect. A large 
proportion of cocker spaniels show almost no avoidance reactions and 

Table 2. — Avoidance and vocal reactions, 5 weeks 






























































































Journal of the National Cancer Institute 



practically all basenjis show them in varying degrees. Inspection of the 
table indicates that the basenji condition is dominant in the first genera- 
tion, as the distributions are quite similar. This is borne out by the 
distributions of the two backcross generations. 

If the data are divided on the basis of a terminal class in which there 
are no reactions, a figure of 0.21 is obtained for the relative distance 
between the backcrosses. This is quite close to the theoretical figure of 
0.25 which would be obtained for two factors with no heterozygosis. 
From the fact that the parent basenjis show considerably less variability 
than the rest of the sample, it may be concluded that there is some 
heterozygosity in the basenji line for this character and the figure of two 
factors must be considered a minimum. 

Aggressive Reactions Toward Human Handlers, 13 to 15 Weeks 

These figures are taken from the same test on which the preceding data 
is based. At 5 weeks of age almost no aggressive reactions are given, but, 
at 13 and 15 weeks, which have been averaged together, biting, chewing, 
pawing and some other aggressive reactions are quite common. As will 
be seen in table 3, the distributions are quite different from the preceding. 
There is a difference between the parent strains but neither appears to be 
a terminal class. The first generation shows greatly increased variability 
compared to the parent strains which might be due to heterozygosity in 
one of the parent strains or might be caused by the threshold phenomenon 
described above, in that the animals have to fall into one of two classes 
and consequently, if the Fi is close to the threshold, it should show in- 
creased variability. The test for the single factor ratio, however, is not 
affected by heterozygosis and would apply in either case. 

Table 3. — Aggression, IS to 15 weeks 






























" i " 





































Vol. 15, No. 3, December 1954 



The relative distance between the backcrosses gives a figure of 0.73 and 
it may be concluded that a minimum of two genetic factors are involved 
and that the situation is one in which the threshold of stimulation is near 
the center of genetic variability. 

Since there is a possibility that the traits of avoidance and aggressive- 
ness may be related, correlation coefficients were calculated for the 
backcross data. It was obvious that in the backcross to the basenji 
there was no correlation, whereas the other backcross showed an obvious 
correlation which would be estimated on the basis of the small numbers 
involved to be at least as high as 0.50. The lack of correlation in the 
basenji backcross can be explained by the fact that the basenji condition 
is dominant for the avoidance scores and, consequently, there is no geneti- 
cally produced variability. It is extremely interesting that the correlation 
in the cocker-spaniel backcross is positive, indicating that an animal which 
shows a high degree of avoidance early in life tends to be more aggressive 
later in life. This means either that the action of the genes on one thresh- 
old of behavior may be quite different on another threshold, or that 
linkage is involved. There is little correlation within either of the two 
pure breeds, indicating that this genetic mechanism is involved chiefly 
in the cross. 

Heart Bate After Weighing 

During the process of weekly checkups the animals are placed upon 
the scale and an attempt is made to keep them quiet for a period of one 
minute, after which the heart rate is taken. A resting heart rate is ob- 
tained if the animal does quiet down, but, since many animals do remain 
active, it probably also reflects the effect of the handler on the behavior 
of the dog. The heart rate is taken for 15 seconds and the figure given in 
the table is the sum of 6 observations taken at 11 to 16 weeks of age. 
When the data are examined (table 4) it will be seen that there is some 

Table 4. — Heart rate after weighing, 11 to 16 weeks 


















""3 ' 








. .... . 






"o" ' 



























Journal of the National Cancer Institute 



indication of a possible difference between the reciprocal crosses, but 
that in general the cocker-spaniel condition appears to be dominant both 
in the Fi and backcross generations. There is a great deal of variability 
in the backcross to the basenjis as one would expect in a segregating group. 
The relative distance between the backcrosses is 0.538, which is very 
close to the theoretical figure for one factor. The data appear to be 
fairly normally distributed and, when the method of analysis of Castle 
and Wright {12) is used, an answer of 1.25 factors is obtained, thus giving 
a check between two methods of calculation. The only difficulty en- 
countered in the application of the Castle and Wright formula was the 
great amount of variability in the segregating generation compared with 
the relative distance between the parent strains. This had to be reduced 
by comparing only variance within litters from corresponding mothers 
from the F x and backcross strains, before any meaningful figure could be 

The Solution of a Simple Barrier Problem 

Puppies at 6 weeks of age which have had no experience with barriers 
are tested on their ability to find their way around an opaque barrier 
beyond which is a goal of food and the experimenter. Successful ex- 
perience is given on the first day and on the second a longer but similar 
barrier is presented . Some individuals find their way around immediately 
and a minimum of three stops or turns (errors) is considered to indicate 
an insightful solution; namely, that the animal is able to organize his 
impressions of the situation to reach a successful conclusion without 
going through random exploration. 

As can be seen from table 5, the great majority of basenjis are successful 
at this, whereas only a few cocker spaniels are as good. The first genera- 
tion appears to be intermediate between two strains. 

Table 5. — Errors, first trial, day 2, 6-week barrier test 












46 & above. . . . 













" i " 






•y • 








































Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS op 

Both backcrosses show a great deal of variability as one might expect 
in segregating strains but, on the average, both are inferior to the cocker 
spaniels. When the distance between the backcrosses is calculated, a 
meaningless figure of —0.064 is obtained. 

No clear genetic analysis can be made from the data and there are 
two possibilities. One is that more than one trait is being measured and 
that these can be separated by more detailed analysis of the data, although 
it is difficult to see how this can be done. The other possibility is that 
more than one trait is associated with the process of organization or 
adaptation going on in the brain of the animal. It would be expected that 
any animal would possess many genetically determined capabilities and 
that in a situation requiring adaptation these would be organized in the 
best way possible. However, in almost any situation there are a variety of 
ways in which capabilities can be organized and there would not neces- 
sarily be a one-to-one relationship between ability and the resultant 
behavior. This, of course, raises the problem of whether additive effects 
will ever be obtained from this kind of adaptive reaction. 

General results. — The data described above are summarized together 
with certain other tests in table 6, and give a small but fairly representative 
sample of the kind of results which are being obtained from the analysis of 
approximately 30 major behavioral tests. As will be seen, the analysis of 
simple behavioral responses in which thresholds are involved gave clear-cut 

Table 6. — Analysis of number of genetic factors affecting each of 9 behavior traits 





Relative position of other ratios 







No. of 
(est. het- 

Avoidance & vocal, 5 wks 

Playful fighting, 13-15 



Heart rate before 

Heart rate after 

Barrier, 6 wks 

Part 2, #1 

Habit, 9 wks 

Part 2, #1 

Barrier, 13 wks 

Total errors 


. 125 











* 940 

*. 367 






1. 240 
-. 148 


-. 312 













'Large difference between reciprocal crosses; like maternal type. 

Journal of the National Cancer Institute 


genetic results with the exception of postural reactions, and it is expected 
that this can be resolved with further analysis. The minimum number of 
factors is two except in one case. It is indicated that a reasonably simple 
genetic mechanism is involved in these traits but that other factors would 
be found in breed populations less limited. The multiple factor hypothesis 
regarding the determination of behavioral traits is thus upheld. 

On the other hand, none of the tests which involve adaptation and 
learning show such clear-cut results. It is possible that more detailed 
analysis will give this result, but it is also likely that the process of adapta- 
tion, in which the animal organizes his sensory and motor capacities to 
meet a new situation, is nonadditive in nature. The reasons for this can be 
seen more clearly in other tests which involve motor organization and 
where the observer can see what the animals are doing. For example, in 
one test the puppies are required to get a dish of food out from under a 
cover and success largely depends on the animal's tendency to move 
objects in some way. Basenjis usually do well by pawing the cover and 
the cocker spaniels by lifting it with the mouth. The Fi generations are 
intermediate in both respects, and usually fail. It is not possible to paw 
and lift objects with the teeth at the same time, and the Fi animals cannot 
add these capacities together. The underlying genetic situation is 
obviously complex and the effects are nonadditive in terms of adaptation. 


In general, the same types of effects which Richter has found resulting 
from selection and domestication in the rat are also found in greater degree 
in the dog. Variability appears to be very great both in anatomical 
characteristics which affect behavior and in types of behavior which have 
no obvious anatomical explanation. Early sexual maturation appears to 
be the rule in all breeds of dogs. While the average fertility is probably 
somewhat decreased from that of the wolf, certain breeds seem to show 
increased fertility. Traits involving wildness and early socialization have 
been greatly modified, and selection for the power to learn a given task 
has proceeded much further in the dog than in the rat. 

From a genetic point of view Richter has done much to demonstrate the 
nature and importance of behavioral and physiological differences between 
wild and tame strains, and to clear up some of the problems raised by 
earlier work done by associates of Castle. There is an extensive literature 
on this subject in the rat and the numerous crossbreeding experiments 
have been reviewed by Scott and Fredericson (16) and Hall (17). How- 
ever, most of these were done before the methods of analyzing multiple 
factor inheritance had been worked out and the results indicated very 
little except that strain differences existed. Usually only very few 
measurements were made so that no idea was obtained as to the nature of 
the traits. Only one of these studies, that made by Dawson (18) on the 
running time of mice which were pushed along an alleyway, gives any 
definite genetic results, and he found that there were between two and 

Vol. 15, No. 3, December 1954 
316263—54 33 

754 proceedings: symposium on 25 years of 

three genetic factors involved in differences between the strains that were 

The work described in this paper on the dog shows that strain differences 
in timidity, and other traits which affect the processes of socialization, 
can be caused by reasonably simple genetic mechanisms, although, if all 
of the traits concerned are considered, many factors may be involved and 
the multiple factor hypothesis is supported. The work has general 
importance in that some process of socialization is found in all social 
animals and upon it depends the successful adaptation of the individual 
to the group and also the successful integration of the entire group. The 
process of socialization is important in human development and its modi- 
fication may result in problems of abnormal behavior [Bowlby (19)]. 
Genetic results on lower animals indicate that such modification might be 
produced more easily in some individuals than in others. 

An interesting negative result was obtained in the hybridization of two 
breeds which presumably have been separated for centuries. It might be 
supposed that modification of the wild type had been achieved by the 
selection of different genes in the two breeds so that the Fi would produce 
a throwback to the wild type. No evidence of this was obtained with 
regard to the traits of wildness or savageness which affect the process of 
socialization. There was evidence of hybrid vigor in some physiological 
characteristics, particularly the amount of milk produced by Fi mothers. 
However, the evidence on behavior indicates that the changes toward 
tameness were probably achieved early in the process of domestication. 
This result also argues for the single ancestry theory which arises from the 
anatomical evidence. 

Several experiments on selection for behavioral differences have been 
done with the domestic rat and positive results achieved in both emotional 
traits, as found by Hall (17), and in maze learning, as found by Tryon (20). 
Searle (21), who studied Tryon's strains with multiple behavioral tests 
came to the conclusion that the chief differences between maze-dull and 
maze-bright rats was that the former were more afraid of the maze than 
the latter, as they perform equally well or better on other types of mazes. 
This work had the result of emphasizing the results of emotional and 
motivational differences as they affect adaptation and the consequent 
importance of testing behavioral differences in many different ways. In 
making crosses between the two strains, Tryon found that the Fi covered 
the distribution of both parent strains with a very wide variance. This 
result is, of course, similar to that found in the test for aggressive behavior 
of the dogs, which suggests that Tryon may have been working with an 
emotional difference in which the threshold of stimulation was near the 
center of the genetic distribution. Unfortunately, no backcrosses were 
made, and so no estimate of the number of factors involved can be made. 

In addition to the crossbreeding studies described, several pure breeds 
of dogs have been extensively tested in a variety of behavioral characters 
[Fuller and Scott (22)]. Kesults indicate that each breed has several in- 
dependent capacities and abilities which affect its ability to learn easily 

Journal of the National Cancer Institute 


the special task for which it has been selected. For example, the Shet- 
land sheep dog has a tendency to form fixed and lasting habits, apparently 
because of great emotional sensitivity. The beagle has the opposite tend- 
ency so that it rarely takes the same path twice, which is of great advan- 
tage in investigating and hunting. Sheep dogs show a considerable 
degree of aggressiveness which results in their tending to chase sheep, 
while beagles tend to be nonaggressive so that dog fights are infrequent 
while they are living in packs. Sheep dogs tend to be indifferent to food 
rewards while beagles are highly motivated by them. Such instances 
could be multiplied indefinitely. On the other hand, there is little evi- 
dence that complicated patterns of behavior will appear spontaneously 
without previous learning. 

Certain traits which make it easy for an animal to learn one type of 
problem may limit its learning ability elsewhere. The sheep dogs which 
are highly adapted to learn commands under close personal supervision, 
do comparatively poorly under kennel conditions and in situations where 
they are required to solve problems independently [Fuller (23)]. The 
aggressiveness of the fox terriers makes them easy to teach to attack small 
game, but when they are raised in litters it is impossible to keep more than 
two or three animals together because of physical injuries inflicted on 
each other. The conclusion may be reached that the exaggeration of a 
behavioral trait by selection may make a particular learning process 
easier, but it may also limit the power of adaptation under other situations. 

A number of interesting problems are raised by nonadditive combina- 
tion of behavioral traits which takes place in adaptive behavior. As 
pointed out above, a single exaggerated trait, such as excessive timidity 
or aggressiveness, may make certain kinds of adaptation impossible. At 
the moment, the most fruitful line of research seems to be to try to dis- 
cover the nature and significance of these components rather than to 
gauge their combined effect upon adaptation. So far we have no evidence 
of general intellectual organizing ability apart from the simple components 
of behavior. If such differences in ability do exist in normal individuals 
they do not appear to be important compared with the simple components. 

These tentative conclusions present several problems to the dog breeder 
who is interested in increasing the natural ability of performance in his 
animals. The over-all performance as measured by the simple speed of 
learning will be a poor measure for selection, although it is the ultimate 
measure of success. Effort should be directed toward discovering what 
the simple components of behavior are and selecting primarily for these. 
If, as in the present experiments, each of these simple elements are affected 
by only one or two genetic factors, it may turn out that the genetic vari- 
ability in any pure breed may be quickly exhausted as seems to have been 
the case where attempts have been made to increase the performance of 
guide dogs by selection. 

Any applications of these results to human problems must be done with 
a great deal of caution. As David and Snyder (24) have pointed out, 
there are good reasons for believing that consistent selection for special 

Vol. 15, No. 3, December 1954 

756 proceedings: symposium on 25 years of 

types of behavior has never been an important factor in human evolution. 
On the other hand, the development of a high degree of social organization 
and consequent protection of the members permits a very wide degree of 
variability between individuals just as similar protection extended to 
domestic animals allows great variability among them. The present 
experiments show that certain kinds of behavioral traits, particularly 
emotional and motivational traits, can be importantly affected by heredity. 
Certain individuals may, as a result, have a very wide range of adaptation 
while others may have a narrower but superior range. The question may 
be asked, do these same conditions exist among human individuals? 

If the answer is in the affirmative our chief practical problems among 
human beings are to find means of recognizing hereditary differences at 
an early age so that the proper environment can be given to the individual 
and he can reach his widest range of adaptability. From the animal ex- 
periments it would be indicated that it would be most profitable to search 
in human beings for the basic emotional and motivational thresholds 
which seem to be the most important determinants of learning and social 
behavior in normal individuals. Since speech is so important a capacity 
in human society the basic motor and emotional capacities which affect 
speech should be another special object of search. 

It should also be remembered that the animal experiments show that 
there is a wide range of adaptability particularly where a learning task is 
complex. There are many different ways in which an individual may 
organize his capacities and equally good results may be achieved by what 
appear to be entirely different individuals. It is to be hoped and expected 
that the recognition and study of basic behavior traits affected by heredity 
will greatly aid the education and behavioral adjustment of the human 


1) Typical changes in behavior of domestic animals include increased 
genetic variability, early sexual maturity, and modification of the social- 
ization process with regard to people (i.e., decreased wildness). In 
addition, breed selection for the ability to learn specialized tasks has been 
carried further in the dog than any other animal. 

2) According to the paleontological evidence the dog was domesticated 
from a single species of wolf in northern Eurasia. 

3) Archeological evidence indicates that domestication took place 
possibly 8,000 years ago and that the use of dogs spread rapidly among 
primitive peoples. 

4) The possession of dogs by semi-isolated human tribes provided a 
favorable condition for genetic change, and each geographical region 
tended to have a different type of dog associated with it. 

5) The social organization of wolves is one which permits easy domesti- 
cation, and there are authentic reports of wolf cubs tamed by rearing apart 
from the mothers. 

Journal of the National Cancer Institute 


6) Five representative breeds were selected for detailed behavioral 
study; beagle, Shetland sheep dog, wire-haired fox terrier, cocker spaniel, 
and African basenji. The last two were chosen for a crossbreeding 

7) These breeds and various hybrid generations have been raised under 
standard enviromental conditions and subjected to a variety of motiva- 
tional, emotional, social, physiological and learning tests at appropriate 

8) A theoretical basis for analysis of multifactorial heredity applied to 
the special case of thresholds of behavioral stimulation is described, 
including a simple test for single factor inheritance. Typical results are 
summarized below. 

9) A measure of avoidance reactions affecting the process of socializa- 
tion toward people gives a minimum estimate of two factors accounting 
for the difference between cockers and basenjis. There is no increase in 
wildness in the Fi, indicating a common genetic mechanism even though 
the two strains have been long separated. 

10) A measure of aggressiveness likewise gives an estimate of two 
factors, but the results indicate a threshold in the center of the genetic 

11) A measure of the resting heart rate gives an estimate of a one- 
factor difference. 

12) A test of performance, in which the puppy has to find its way 
around a barrier on the basis of previous experience, does not give a clear 
picture of the genetic mechanism. 

13) In general, measures of differences in the threshold of stimulation in 
simple behavior patterns give reasonably simple genetic results, although 
usually with more than one factor involved. 

14) Measures of complex adaptation do not give simple genetic results, 
and it is suggested that this may result from the fact that adaptation con- 
sists of organizing the many capacities of the individual, and that this can 
take place in a variety of ways to produce the same result. 

15) Capacities which increase the ability to learn a special situation may 
limit the range of adaptability in other situations. 

16) With regard to human applications, it is concluded that the genetic 
situations in the two species are quite different, but that the dog results 
give an estimate as to how great genetic differences between individuals 
may be. The most profitable line of human research would appear to be 
the search for early differences in basic emotional and motivational traits, 
together with the types of environment which give the most desirable 
expression of these traits. 


{1) Richter, C. P.: Domestication of the Norway rat and its implication for the 
study of genetics in man. Am. J. Human Genet. 4: 273-285, 1952. 

{2) Young, S. P., and Goldman, E. A.: The Wolves of North America. Wash- 
ington, D. C., American Wildlife Institute, 1944. 

(3) Little, C. C. and DuBuis, E.: Private communication, 1954. 

(4) Matthew, W. D.: The phylogeny of dogs. J. Mamm. 11: 117-138, 1930. 

Vol. 15, No. 3, December 3951 

758 proceedings: symposium 

(5) Allen, G. M.: Dogs of the American aborigines. Bull. Mus. Comp. Zool. 

Harvard 63: 431-517, 1920. 

(6) Degerbol, M.: tlber prahistorische danische Hunde. Vidensk. Meddel. 

Dansk. Naturhist. For. Kobenhavn 84: 17-72, 1927. 

(7) Haag, W. G.: An osteometric analysis of some aboriginal dogs. Univ. of 

Kentucky Rep. Anthropol. 7: 107-264, 1948. 

(8) Murie. A.: The Wolves of Mount McKinley. U. S. D. I. Fauna Ser., No. 5, 

Washington, D. C., Government Printing Office, 1944. 

(9) Scott, J. P.: The social behavior of dogs and wolves: an illustration of socio- 

biological systematics. Ann. New York Acad. Sc. 51: 1009-1021, 1950. 

(10) Williams, V. T.: Basenjis, the Barkless Dogs (revised ed.). London, Wat- 

moughs, 1954. 

(11) Scott, J. P., and Fuller, J. L.: Manual of dog testing techniques (mimeo- 

graphed). Roscoe B. Jackson Memorial Laboratory, Bar Harbor, 1950. 

(12) Castle, W. E.: An improved method of estimating the number of genetic factors 

concerned in cases of blending inheritance. Science 54: 223, 1921. 
(18) Wright, S.: The results of crosses between inbred strains of guinea pigs, differ- 
ing in number of digits. Genetics 19: 537-551, 1934. 

(14) * The genetics of quantitative variability. In Quantitative Inheritance. 

London Agricultural Res. Council, 1952. 

(15) Scott, J. P.: Biomathematical problems in animal behavior. In Statistics and 

Mathematics in Biology. Ames, Iowa, Iowa State College Press, 1954. 

(16) Scott, J. P., and Fredericson, E.: The causes of fighting in mice and rats. 

Physiol. Zool. 24: 273-309, 1951. 

(17) Hall, C. S.: The genetics of behavior. In Handbook of Experimental Psy- 

chology, New York, Wiley, 1951. 

(18) Dawson, W. M.: Inheritance of wildness and tameness in mice. Genetics 17: 

296-326, 1932. 

(19) Bowlby, J.: Maternal care and mental health. World Hlth. Org., Geneva, 1951. 

(20) Trton, R. C: Individual differences. In Comparative Psychology. New 

York, Prentice-Hall, 1942. 

(21) Searle, L. V.: The organization of hereditary maze-brightness and maze- 

dullness. Genetic Psychol. Mono. 39: 279-326, 1949. 

(22) Fuller, J. L., and Scott, J. P.: Heredity and learning ability in infra-human 

mammals. Eugenics Quart. 1: 28-43, 1954. 

(23) Fuller, J. L.: Cross-sectional and longitudinal studies of adjustive behavior in 

dogs. Ann. New York Acad. Sc. 56: 214-224, 1953. 

(24) David, P. R., and Snyder, L. H.: Genetic variability and human behavior. In 

Social Psychology at the Crossroads. New York, Harper, 1951. 

The Effects of Selection and Domesti- 
cation on Man x 

Laurence H. Snyder, The University of Okla- 
homa. Norman. Okla. 

The very title which has been assigned to me presents an enigma: Can 
man properly be said to be domesticated, or even to have been subjected 
to the process of domestication? To reclaim from the wild state, to cicu- 
rate or domesticate, logically implies a reclaimer — a domesticator. Man 
himself plays this role in the usual examples of reclamation of animals and 
plants ; can he then wear two hats in regard to his own species? Can he be 
both domesticator and domes ticatee? It has been authoritatively argued 
that he can (1), and indeed our previous speaker has both here and else- 
where (#) clearly indicated that he views man as having passed through 
the process of becoming domesticated. 

For the purposes of this discussion we shall perhaps make more progress 
if we consider the self-domestication of man as synonymous with the 
process of civilization. This process has been marked by two outstanding 
phenomena: first, the creation of an environment of ever-increasing safety 
and security; and second, the development of culture. It goes without 
saying that I use the word culture in regard to the intellectual aspects of 
civilization, and not at all in the arrogant and restricted meaning by which 
it is often abused, 

Natural selection in feral man probably favored strength, aggressiveness 
and freedom from physical and mental defect. Primitive environments 
presented the necessity of constant struggle for food, shelter and mates, 
and of recurring defense against enemies. But man possessed certain 
specializations, which, though few in number, were unique. These 
specializations enabled him to rearrange his environment in such a way as 
to provide ever more efficiently the essentials and even the comforts of 
life, and to minimize the imminent dangers, thus allowing the development 
of high attainments in art, religion, statecraft, law and science — in short, 
culture. Concurrently the selective advantages of sheer strength and 
aggressiveness probably declined. 

The specializations which have permitted the growth of cultural patterns 
include the fully upright position and its concomitants, the differentiation 
of feet and hands ; and the exceptional size of the brain, with the develop- 
ment of the areas involved in communication, abstraction and symbolizing. 
Aside from these, man is quite generalized, and lacks the specializations 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 29, 1954. 


Journal of the National Caneer Institute, Vol. 15, No. 3, December 1954 

760 proceedings: SYMPOSIUM ON 25 YEARS of 

which are characteristic of such mammals as bats, whales, giraffes, sloths 
or elephants. 

Human beings are most uniform in those characters which they share 
with the greatest variety of other animals, and differ one from another 
more and more in regard to traits of increasing superficial ty (3). Thus, 
the fundamental chordate pattern is a remarkably constant characteristic 
of human beings; the ossified skeleton is somewhat more subject to varia- 
tion; the more strictly mammalian characters involving skin, hair, mam- 
mary glands and control of body temperature are quite subject to modifi- 
cation. Mutations involving the more fundamental, deep-seated chordate 
traits are quite likely to be lethal, but changes in the more superficial 
characters may very well be viable, especially as those which might 
otherwise be detrimental can often be compensated for by man's ingenuity 
in controlling his environment. 

It is difficult to specify which, if any, of the mutations causing vari- 
ability in man have attained their present population frequencies as a 
direct result of the process of domestication. It is even more difficult 
to specify which, if any, of the human traits other than physical peculiar- 
ities are in fact the result of genetic determiners at all (4, 5). Kroeber (1) 
lists the following anatomical characteristics as being directly associated 
with the domestication of man: the long hair on the head; the near-hair- 
lessness of the body; curly hair, woolly hair, blond hair; blue eyes and 
fair skin. To these Richter (2) has added the possibility that, in line with 
experimental data from rats, the adrenal glands of civilized man may have 
decreased in functional importance, while the hormonal activity of the 
gonads may have increased in salience. Such changes, if substantiated, 
might well be expected to exert significant effects on behavior. 

In the course of man's domestication the usual evolutionary processes 
have acted upon him. Through geographic or cultural isolation and the 
concomitant effects of selection, inbreeding, mutation and genetic drift, 
various populations have become more or less differentiated one from the 
other in respect to readily recognizable physical traits, such as skin color, 
hair form, head shape and occasionally stature. Whether such ethnic 
groups differ significantly in regard to mental traits we shall inquire into 

Within ethnic groups are found smaller, more or less self-contained 
breeding units, which we call isolates (6). These isolates are delimited 
by social class, religious affiliations, habitat, and other cultural and 
geographical isolating mechanisms. They may in some instances be 
distinguished by a way of life. Nomadism and agriculture, for example, 
require for successful pursuance quite divergent traits of temperament, 
and Huxley (7) alleges that the followers of these modes of existence early 
became differentiated in many such qualities. 

In regard to sets of alleles whose existence in man has been well estab- 
lished, the proportions of the various alleles can be shown to vary not 
only from one ethnic group to another, but from isolate to isolate (8). 
Even within isolates is found residual genetic variation, some of it due 

Journal of the National Cancer Institute 


to single, major gene substitutions, and some to polygenic heredity (9). 
The dynamics of major gene transmission present implications for human 
genetics quite different from those evolving from polygenic systems. 
These implications have been analyzed in some detail elsewhere (10, 9), 
but a few of the more relevant consequences may be discussed here. The 
vast majority of the discontinuities which are dependent on single gene 
substitutions are pathological in nature. Because of the viability impair- 
ment connected with them, the incidences of such anomalies and diatheses, 
and thus of the genes responsible for them, are low. Under the restraining 
influence of natural selection, only a very few such traits have reached 
population incidences above one in 10,000 and the vast majority are much 
rarer than this. 

Conditions of modern civilization, however, have resulted in a relaxation 
of selection against some genes, and in the subjection of other genes to new 
selective processes. Some of the potential outcomes of these man-made 
shifts in evolutionary trends warrant thoughtful scrutiny and deserve 
careful analysis. 

One of the many consequences of man's self-domestication has been 
the phenomenal progress of the science of medicine. Not only has this 
progress resulted in the control of numerous environmentally conditioned 
diseases, but it has led to the alleviation of various genetically determined 
disorders and anomalies (11) . The result is that some types of individuals 
who formerly were eliminated before reproducing, or before completing 
their families, are now enabled to live out a more normal span of life and 
are given the opportunity of more normal reproduction. The presumed 
implications of these facts have aroused concern in certain quarters 
(12, 13). The concern is, in my opinion, unjustified, and I should like to 
point out once more the underlying fallacy, to which attention has else- 
where been directed (14). 

Throughout the long course of evolutionary history, living organisms 
have been subject to continuing change. One of the principal causes of 
genetic change is mutation. An important consequence of mutation is 
that the mutated gene reproduces itself just as faithfully in its new mo- 
lecular arrangement as the unmutated gene did in its original chemical 

It is a familiar observation that those mutant genes which produce 
conspicuous effects commonly behave as recessive genes, although these 
same genes may also produce less noticeable effects which, by special 
techniques, are detectable in the heterozygote (15). Consequently the 
occurrence of a new mutation is not very likely to be noticed until two 
heterozygous individuals mate and produce a clearly affected homozygous 
offspring. Mutation at a particular locus may occur in more than one 
gamete or in more than one individual, and in fact the phenomenon is 
characteristically a recurrent event. Thus, despite the infrequent 
incidence of origin of any particular mutant gene by mutation, it will 
tend gradually to accumulate in the species. The tendency to accumulate 
may be opposed by natural selection in those instances in which the 

Vol. 15, No. 3, December 1954 

762 proceedings: symposium on 25 years of 

effects of the mutant gene impair to any degree the average reproductive 
rate of those who possess the gene. 

Concomitant with long ages of mutation and selection, living organisms 
are characteristically highly adapted and delicately adjusted mechanisms. 
As a result, fortuitous new mutations, except those involving only super- 
ficial traits, are quite likely to give rise to deleterious, even lethal, effects. 
Among innumerable fortuitous mutations over the ages, however, some 
few have been favorable or beneficial, and from the selective advantages 
conferred upon genes of this nature has come evolutionary progress. 

Let us return then to the direct consideration of the concern which 
is being commonly expressed these days over the possibility that the 
accumulation even of spontaneous mutations in man may eventually 
constitute a serious threat to the public health. The argument runs 
somewhat as follows: Since mutation is a recurrent phenomenon, mutant 
alleles would gradually accumulate in the human species were it not for 
the fact that many mutations are harmful. This fact permits the tendency 
to accumulate to be opposed by the process of natural selection, thus 
depressing the frequency of harmful genes to an equilibrium value at 
which the rate of elimination is balanced by the rate at which the mutant 
gene arises by fresh mutation in any generation. 

Now the progress of civilization, and especially of medicine (it is 
argued), has succeeded in ameliorating the effects of many harmful genes, 
so that selection against these deleterious genes has been relaxed, thus 
permitting their frequencies to attain higher and higher equilibrium 
levels. Contemplation of the long-range results of further successes in 
ameliorating the effects of more and more undesirable genes presents an 
alarming prospect to proponents of the argument as I have reported it. 

The essential fallacy of the argument (14) consists in the application of 
the epithets "deleterious/ ' "harmful," or "undesirable" to the mutant 
genes themselves rather than to their effects. If, through modern 
medical, social or economic progress, selection has been relaxed against 
any gene with harmful effects, this relaxation has been accomplished only 
because the medical, social or economic agencies have provided environ- 
ments in which the effects of the genes are rendered less harmful or quite 

As to those genes that persist in producing detrimental effects in all 
known environments and despite all attempts at therapy, selection against 
them remains today as severe and effective as ever. And selection will 
again begin to operate against any gene for which it has been relaxed if 
the burden of providing the necessary therapeutic conditions begins to 
outweigh the social value of providing them. It seems reasonable to 
presume, however, that medical and social advances will continue to be 
made with ever-increasing efficiency, and that therapeutic or preventive 
measures which may now seem burdensome will be continuously improved 
and will become ever more simple, natural and acceptable. 

One need only consider some of the many commonplace procedures 
which probably were at one time burdensome but which are now com- 

Journal of the National Cancer Institute 


fortably incorporated into our modes of life. Thus we compensate for 
our loss of natural ability to control adequately the temperature of our 
bodies by providing ourselves with varying degrees of clothing and with 
temperature-controlled dwellings. We add vitamins to our diets, and 
take hormones when necessary. We successfully feed and rear infants 
in the absence of human milk. We wear glasses when indicated, we see 
our dentists at least twice a year, and we are reasonably happy under 
these restrictive derivatives of civilized existence. I see no reason to 
dread the genetic effects of further advances in medical or social science. 

Let us turn our attention now to the problem of the extent to which, 
during, and as a result of, the domesticatory process, man's behavior may 
be attributable to genetic factors. It is clear from the foregoing presen- 
tations in this symposium that a genetic basis for behavior is to be found 
in some domestic animals, and that the process of domestication has 
exerted selective action on the gene constellations involved. Indeed the 
existence of polygenes and possibly of major genes which affect tempera- 
ment, motivation, social activity and other behavioral responses in experi- 
mental animals has been well established by various workers (2, 16-20). 

There may well be in the human species polygenic systems basic to the 
expression of organic drives analogous to those in animals, but it is 
probable that as determinants of social roles, attitudes and behavior, 
genetic factors are of limited significance in man. The rationale for this 
point of view has been presented in detail in an earlier publication (10). 
The salient points in the argument may well be discussed here anew from 
the standpoint of domestication. 

First, there may be some doubt as to the extent to which behavior in 
animals should be equated with that in man. To admit "the obvious 
and close analogy between human and rodent types of boldness" (21), 
for example, is to ignore the striking qualitative differences between the 
significance of organic drives for animals whose interindividual relation- 
ships are on a physiological or biosocial level, and their significance for 
human beings, whose interpersonal relationships are on a psychosocial 
level (22, 23). Genetic variables affecting behavior in animals may fairly 
be postulated as exerting their effects on neurophysiologic mechanisms. 
While certain severe (but rare) neurologic aberrations in man also are 
undoubtedly due to genetic interference with neurophysiological mecha- 
nisms, the overwhelming bulk of man's behavioral variability must surely 
occur as a result of functional disturbances in the human cerebral cortex, 
to which I have already pointed as a specialization unique in man. 

Undomesticated animals and plants have developed evolutionarily by 
the adapting of their characteristics to existing environments. Domesti- 
cated organisms have had their characteristics adapted by human manip- 
ulation to environments which were themselves at the same time being 
adapted by man to the changing traits of the organisms. Husbandry is 
a complex art. Man himself, however, has become adapted evolutionarily, 
by consciously and purposefully altering his own environment by means 
of his own inventions, to meet existing genotypes. Moreover, man's 

Vol. 15, No. 3, December 1954 

764 proceedings: symposium on 25 tears of 

ability to reason, to communicate, and to record his thoughts in writing 
has made the environmental improvement a cumulative process. The 
result is all too apparently the gradual outstripping of biological inherit- 
ance by social inheritance, or tradition, as a major social force. 

Various writers have interpreted in widely varying ways the prospect of 
the replacement of the mechanism of the chromosome and the gene by the 
forces of social organization as the major determinant of human behavior. 
Some, like Orwell (24) and Seidenberg (25), take pessimistic views indeed. 
In clear, sure strokes Seidenberg paints an engrossing verbal picture of 
the long conflict between instinct and intelligence — the one dominant 
because of its evolutionary priority, the other offering a challenge by 
virtue of its inherently cumulative power. He sees, as the result of the 
usurpation of the dominant role by tradition, the gradual approaching of 
fixed perimeters, of a crystallized status of man within compelling forms 
of organized procedures. In the ultimate state of crystallization to 
which the principle of organization leads, Seidenberg predicts that con- 
sciousness will have accomplished its task, leaving mankind sealed, as it 
were, within patterns of frigid and unalterable perfection. 

The subject is developed logically and inexorably, with frequent well- 
chosen documentation. The ultimate triumph of organization over 
individuality is presented as inevitable, and is a frightening prospect 
indeed. The terrible fascination of the author's logic, which can be 
experienced only by reading the volume itself, holds the interest of the 
reader to the implacable end. 

In contrast to the foregoing views, other writers (10, 16, 27) take the 
position that, although man's ability to control and alter his surroundings 
has indeed brought about more and more uniformity of the physical 
environment, it has at the same time resulted in greater and greater 
heterogeneity of the social environment. Not only are human social 
environments different from place to place, but they have in the course 
of the history of mankind been extremely varied, and in terms of evo- 
lutionary time have succeeded each other with considerable rapidity. 
As a consequence, man has been subjected to much of the external vari- 
ability required for molding the frequencies of such alleles as might be 
involved in various aspects of behavior. 

Of the two obvious possible responses, namely, the selection of ever 
more precise specializations which would fit him to cope with specific 
environmental situations, and the development of increasingly plastic 
responsiveness of any of a variety of environmental situations, man 
apparently followed the second trail (27). It is easy to accept the sug- 
gestion (26) that the one strong selective pressure to which mankind has 
been continuously and unremittingly subjected during his period of 
domestication and in all social environments is selection for educability — 
for the capacity to benefit from experience and reasoning. Just as one 
of man's striking physical assets is the relatively generalized human hand, 
which can be put to a variety of uses, so plasticity of responsiveness, 
permitting adjustment to a variety of situations, has been characteristic 
of man's psychic evolution. 

Journal of the National Caneer Institute 


It seems highly probable that in all human social situations plasticity 
of response, and emotional and temperamental resilience, have been of 
sufficient value as to have been at a selective advantage. If this be true, 
it is highly improbable that any particular population ever developed 
significant genetic differentiation in respect to specific response patterns, 
temperaments, personality types, or intellectual capacities. In other 
words, despite the exceptional diversity of genotypes among human 
beings (8), the plasticity of individual response to the social environment 
is even more noteworthy. The dynamics of polygenic population genetics 
lend strong support to such a view (10). Thus we arrive logically at our 
original conclusion that, as determiners of specific and temporal social 
roles and attitudes, genetic factors are, on the whole, of limited signifi- 

In the long-range outlook, however, the genetic endowment takes its 
place as co-equal with the social millieu in the forging of societies. From 
the vantage point of history one may observe the rhythm of the growth 
and decline of civilizations. And perhaps it is, after all, the historian 
who can successfully bridge the gap between the viewpoints of such 
writers as Siedenberg and Orwell on the one hand, and Dobzhansky and 
Montagu on the other. Toynbee, for example (28), develops the con- 
cept of "etherialization," an overcoming of material obstacles, leading to 
the release of the energies of society to make responses to challenges 
which are henceforth internal rather than external, spiritual rather than 
material. Although Toynbee writes with theistic overtones, there re- 
mains, when these are stripped away, an important basic historical truth 
in the concept. 

Throughout history the process of cultural acquisition has resulted in 
bursts of growth and development of societies, alternating with their 
crystallization and ultimate degeneration. Toynbee epitomizes this by 
concluding that, just as differentiation is the mark of growth, so standard- 
ization is the mark of disintegration. But while history thus furnishes 
Seidenberg with some support, in that standardization has indeed been 
the culmination of many societies in the past and may well occur again 
in the future, history provides at the same time the clue to the answer 
to Seidenberg's dilemma. It lies in the very fact that whenever standard- 
ization of thought and social organization has from time to time threatened 
the existence of civilization, new lines of behavior and of development 
have sprung up out of the tremendous potential of variability of response 
inherent in the human species. And with the increasing ability of social 
psychology to provide effective techniques for the management of inter- 
personal and intergroup relations (29, 80), we may look forward hopefully 
to the ever more efficient use of the really enormous constructive poten- 
tialities which are biologically characteristic of mankind as it exists today. 

The preceding considerations carry with them the clear implication 
that there is at present no great danger that the civilized world is facing 
genetic deterioration in regard to intelligence. I suspect that I must not 
end this discussion without presenting my reasons for this belief, since 

Vol. 15, No. 3, December 1954 

766 proceedings: symposium on 26 years op 

there are widely publicized allegations to the contrary (81-38). It is 
claimed that fertility differentials are today of such nature that they 
lead to a decline in the average intelligence of the populations of Great 
Britain and America, at least, of from one to four I. Q. points per gener- 
ation. If the reduction were considered to be merely in terms of pheno- 
typic manifestation there might seem less cause for alarm, but those who 
allege the decline see it as involving genotypic deterioration as well. The 
current argument for the thesis is the apparent existence of a negative 
correlation between I. Q. and fertility — correlation said to be at least 
partly independent of social and economic status. This argument was 
adopted upon the realization of the untenability of the older premises 
involving the reproductive rates in various socio-economic classes and 
their supposed relation to I. Q. levels (84, 85). 

Careful scrutiny of the argument, however, indicates that the actual 
data on which the belief in the negative correlation between I. Q. and 
fertility is based merely indicate an inverse relationship between the test 
intelligence of children and the number of their brothers and sisters (10). 
This relationship does not necessarily imply a similar negative correlation 
between intelligence and fertility, since the lower grade mental defectives, 
whatever the size of their sibships, are generally not themselves fertile. 
Proof that differences in fertility actually permit phenotypic selection 
against high intelligence must rest on the demonstration that any negative 
regression of fertility on I. Q. over one portion of the range of intelligence 
is not compensated for by a positive regression over another part of the 

Moreover, any statement regarding phenotypic selection based on 
reproductive differentials must take into account the demonstrable 
dependence of I. Q. development upon environmental factors, and especial- 
ly on such factors as are associated with educational facilities and cultural 
stimuli (6). Differentials in I. Q. can usually be demonstrated between 
various socio-economic levels, between rural and urban populations, and 
between regions in which public school expenditures are at variance. 
But there is increasing evidence that changes in these environmental 
situations are reflected in changes in the I. Q. In one isolated population 
in a depressed area, for example, the very low average intelligence test 
scores recorded in 1930 were found to be more than ten points higher for 
comparable age groups following a decade of considerable improvement 
in social and economic conditions and in educational opportunities (86). 
The results of the second series of tests in the comprehensive survey which 
has been conducted in Scotland for more than 20 years indicate that the 
average test-intelligence level in the population has certainly not declined, 
and may have improved slightly (87). 

I do not doubt that fertility differentials can be such as to produce 
phenotypic changes in population I. Q., perhaps at a rapid rate, and in 
either direction. But this admission is not to concede that genotypic 
changes are necessarily involved. The belief that they are involved 
appears to rest ultimately on the observable correlation in I. Q. between 

Journal of the National Cancer Institute 


parent and child: a correlation which may very easily be confused with 

It is possible, for example (38), that the deliberate choice of early 
marriage and family responsibilities reduces the chance of becoming 
eminent or even of attaining the full potential of test intelligence, without 
in the least lowering the genetic worth of the individual. The limiting 
factors which determine human fertility are probably, over the greater 
part of the range of intelligence, social rather than biologic. Moreover, 
the factors in the social environment which discourage fertility are in 
general those which tend to permit the development of the full potential 
of the I. Q., whereas the conditions which encourage fertility are broadly 
the same as, or intimately related to, those which tend to depress the 
phenotypic development of high-test intelligence in many, if not all 
genotypes (10). 

It must be kept in mind, moreover, that the genetic background for 
intelligence (and nothing I have said is to be inferred as denying such a 
background) is undoubtedly polygenic in nature. Phenotypic differ- 
entiation in polygenic traits is less likely to result from genetic drift than 
in characters contingent on major genes, since the effects of individual 
polygenes are in large part mutually interchangeable (39, 9) . Neither is 
it probable, as has already been pointed out in regard to behavioral 
responses, that natural selection has resulted in the complete fixation of 
different constellations of genes for intelligence in different populations. 

If the viewpoints just sketched have any validity (and I believe they 
have), it is obvious that fertility differentials will under present cultural 
conditions have minimal effect on the frequencies of genotypes which 
may conceivably be involved in determining innate potentialities for 
intelligence-test performance. 

Looking to the future, we must in all sincerity pin our hopes on the 
management of social relationships. Changes in human social organiza- 
tion have taken place at unbelievably rapid rates in the past, and probably 
will continue to do so in the future. On the other hand, man's biological 
evolution has been, and will surely continue to be, very slow indeed. 
Little or no alteration in man has occurred anatomically in the last half 
million years, and there is little to indicate that intelligence has greatly 
changed. I doubt that the development of the atomic bomb is the mani- 
festation of any greater innate mental capacity than such creative efforts, 
largely buried in the vaults of antiquity, as the invention and employment 
of the wheel or the bow and arrow. Let us wish all success to our col- 
leagues, the social scientists, in their efforts to enable man to use more 
constructively the enormous latent biological potentialities which exist in 
all peoples everywhere. 


(1) Kroebeb, A. L.: Anthropology. New York, Harcourt, Brace and Co., 1948. 

(2) Richter, C. P.: Domestication of the Norway rat and its implication for the 

study of genetics in man. Am. J. Human Genet. 4: 273-285, 1952. 
(S) Snyder, L. H.: The genetic approach to human individuality. Sci. Month. 68: 
165, 1949. 

Vol. 15, No. 3, December 1934 

768 proceedings: symposium on 25 years of 

(4) David, P. R., and Snyder, L. H.: Genetics and disease. Proc. Second Nat. 

Cancer Conf. New York, American Cancer Society, 1954, p. 1128. 

(5) : Principles of human genetics. In Genetics and the Inheritance of 

Integrated Neurological and Psychiatric Patterns (Hooker, D., ed.). Res. 
Publ. Assn. Res. Nerv. and Ment. Dis., No. 33, 1954. 

(6) Snyder, L. H.: The Principles of Heredity, 4th ed. Boston, D. C. Heath and 

Co., 1951. 

(7) Huxley, J.: Genetics, evolution and human destiny. In Genetics in the 20th 

Century (Dunn, L. C, ed.). New York, The Macmillan Co., 1951. 

(8) Snyder, L. H.: Principles of gene distribution in human populations. Yale J. 

Biol. & Med. 19: 817-833, 1947. 

(9) : Old and new pathways in human genetics. Am. J. Human Genet. 3: 

1-16, 1951. 

(10) David, P. R., and Snyder, L. H.: Genetic variability and human behavior. 

In Social Psychology at the Crossroads (Rohrer, J., and Sherif, M., eds.). New 
York, Harper and Bros., 1951. 

(11) Snyder, L. H.: Frontiers in genetics. In Frontiers in Medicine: The March 

of Medicine 1950. New York, Columbia University Press, 1951. 

(12) Cook, R. C: Lethal genes a factor in fertility. Eugenics News 38: 49-55,1953. 
(IS) Muller, H. J.: Our load of mutations. Am. J. Human Genet. 2: 111-176, 


(14) Snyder, L. H., and David, P. R.: Heredity and preventive medicine. In Text- 

book of Preventive Medicine (Leavell, H., and Clark, E. Gurney, eds.). New 
York, McGraw-Hill Book Co., Inc., 1953. 

(15) Neel, J. V.: The detection of genetic carriers of inherited disease. In Clinical 

Genetics (Sorsby, A., ed.). London, Butterworth and Co., Ltd., 1953. 

(16) Fuller, J. L., and Scott, J. P.: I. Heredity and learning ability in infrahuman 

mammals. Eugenics Quart. 1: 28-43, 1954. 

(17) Hall, S. C: The genetics of behavior. In Handbook of Experimental Psy- 

chology (Stephens, S. S., ed.). New York, John Wiley, 1951. 

(18) Heron, W. T.: The inheritance of brightness and dullness in maze learning 

ability in the rat. J. Genet. Psychol. 59: 41-49, 1941. 

(19) Scott, J. P.: Genetic differences in the social behavior of inbred strains of mice. 

J. Hered. 33: 11, 1942. 

(20) Tryon, R. C: Individual differences. In Comparative Psychology (Moss, F. A., 

ed.). New York, Prentice Hall, Inc., 1946. 

(21) Murphy, G.: Genetic and social significance of differential fertility; review of 

relevant research on inheritance of metal traits. Milbank Mem. Fund Quart. 
25: 373-382, 1947. 

(22) Schneirla, T. C: Problems in the biopsy chology of social organization. J. 

Abn. and Soc. Psychol. 41: 385-402, 1946. 

(28) : The "levels" concept. In Social Psychology at the Crossroads. Rohrer, 

J., and Sherif, M., eds.). New York, Harper and Bros., 1951. 

(24) Orwell, G.: "1984." New York, Harcourt, Brace and Co., 1949. 

(25) Seidenberg, R.: Posthistoric Man. Chapel Hill, Univ. North Carolina Press, 


(26) Dobzhansky, T.: The genetic nature of differences among men. In Evolution- 

ary Thought in America (Persons, Stow, ed.) . New Haven, Yale Univ. Press, 

(27) Dobzhansky, T., and Ashley Montagu, M. F.: Natural selection and the 

mental capacities of mankind. Science 105: 587-590, 1947. 

(28) Toynbee, A. J.: A Study of History. New York and London, The Oxford 

Univ. Press, 1947. 

(29) Rohrer, J., and Sherif, M.: Social psychology at the crossroads. In The Univ. 

of Oklahoma Lect. in Soc. Psych. New York, Harper and Bros., 1951. 
(SO) Sherif, M., and Wilson, M. O., eds.: Group relations at the crossroads. The 
Univ. of Oklahoma Lect. in Soc. Psychol. New York, Harper and Bros., 1953. 

Journal of the National Cancer Institute 


(31) Burt, C: Intelligence and fertility. Occasional Papers on Eugenics. London, 

Hamilton Hamish Medical Books, 1946. 

(32) Cattell, R. B.: Effects of human fertility on the distribution of intelligence and 

culture. 39th Yearbook, Nat. Soc. Stud. Educ, Part 1: 221, 1940. 

(33) Thompson, G.: The trend of national intelligence. Occasional Papers on 

Eugenics. London, Hamilton Hamish Medical Books, 1947. 

(34) Osborn, F.: Preface to Eugenics. New York, Harper and Bros., 1940. 

(35) Snyder, L. H.: The genetic and biologic bases of mental disorders. In Mental 

Health. A. A. A. S. Symposium No. 9, 1939. 

(36) Wheeler, L. R.: A comparative study of the intelligence of East Tennessee 

mountain children. J. Educ. Psychol. 33: 321-334, 1942. 

(37) Mental Survey Committee (G. H. Thompson, Chm.). The Trend of Scottish 

Intelligence. London, Univ. London Press, 1949. 

(38) Price, B.: Reference data on Moscow families of 1935. Psychol. Bull. 35: 696- 

697, 1938. 

(39) Mather, K.: Polygenic inheritance. In Clinical Genetics (Sorsby, A., ed.). 

London. Butterworth and Co., Ltd., 1953. 

Vol. 15, No. 3, December 1954 

316263—54 34 

Discussion : Session VI 
Dr. John L. Fuller, Roscoe B. Jackson Memorial Laboratory, Bar Harbor, Maine 

It is obvious from the high quality of these three papers that the genetic method can 
make a definite contribution to the broad area of behavioral sciences. I shall comment 
only upon certain special aspects of each paper, since a full evaluation would require 
much thought and time. 

Dr. Richter presented an account of domestication of the rat in which he placed 
great emphasis upon changes in the relative importance of the gonads and adrenal 
glands. Certainly much of what he said probably applies to other domestic species. 
However, selection under domestication is not always in the direction of docility. 
Certain terrier breeds have been selected for aggressiveness to a degree that appears to 
he detrimental to the establishment of a stable social organization when the dogs are 
living in small naturalistic groups. Would the same adrenal changes occur here, as in 
a strain selected for docility? 

I also wish to express some caution concerning the idea that the strain differences in 
behavior are a direct consequence of endocrine-gland differences. Recently we com- 
pared adrenalectomized, intact, and cortisone-injected members of two strains of mice 
which were rather sharply different in activity, emotionality, and the frequency of 
occurrence of several specific behavior patterns. Drastic changes in the hormones did 
not modify the strain differences in behavior. I believe that we can profitably go for- 
ward in this kind of study viewing the brain as a target organ, and comparing target 
organ-response to hormones in wild and domestic races, and in behaviorally different 
domestic races. 

Dr. Scott's paper deals with two major topics — the history of domestic dogs, and an 
experiment designed to learn about the genetic nature of the differences in breed be- 
havior which have been established by selection. Scott's conclusion that selection has 
not fundamentally altered the behavior patterns of the wolflike ancestor of dogs has 
obvious implications for that civilized primate, man. 

It is impossible to adequately discuss the detailed genetic results which Dr. Scott 
has presented this afternoon without more time than I have available. He deserves 
a great deal of credit for applying a variety of genetic and psychological hypothetical 
models to his data. I should like to point out that the value of the data does not 
depend upon the ultimate verification of each and every hypothesis which he has adopted. 
My guess is that several different approaches will be necessary before we can judge the 
most fruitful model for a genetics of behavior. Scott's work is a pioneering effort in 
this field. 

Finally in considering Dr. Snyder's paper, I should like to comment on his hopeful 
evaluation of man's genetic future. I believe he is correct in his appraisal, but I want 
to stress the fact that this sanguine outlook is not based upon a belief that man is 
superior to biological forces. Biological and social conditions at present are not 
eliminating valuable genotypes while preserving poor ones, but I suspect that deliberate 
selection of man could lead to races of diverse intellectual potentiality. No one is 
seriously proposing this. I simply mention it to emphasize the fact that human behavior, 
like that of the dog and rat, has a biological substratum. 

Students of heredity and of behavior have sometimes seemed to speak in different 
languages. These papers indicate that a new synthesis is in the making. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Session VII. Genetic Techniques in the 
Study of Cancer : New Approaches 

Chairman, Dr. Howard B. Andervont, 
Chief j Laboratory of Biology, National Cancer 
Institute, Bethesda, Md.; Scientific Director, Roscoe 
B. Jackson Memorial Laboratory 

Speaker: Dr. Walter E. Heston 
Localization of Gene Action in the Causation of Lung and Mammary 

Gland Tumors in Mice 
Discusser: Dr. L. M. Dmochowski 

Speaker: Miss Margaret M. Dickie 

The Use of F t Hybrid and Backcross Generations to Reveal New and/or 

Uncommon Tumor Types 
Discusser: Dr. Morris Barrett* 

Speaker: Dr. Elizabeth Fekete and Dr. Allen B. Griffen 

Signiiicance of Recent Developments in Nuclear Cytology and Cytogenetics 

of the Mouse 
Discusser: Dr. Donald F. Jones 

'Discussion not submitted. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Introduction : Session VII 
Dr. H. B. Andervont, Chairman 

Previous sessions of this symposium presented progress in mammalian genetics since 
the founding of the Roscoe B. Jackson Memorial Laboratory in 1929. This morning 
we shall hear how the basic contributions of these geneticists are assisting in the fight 
against cancer. Thus, our meeting will demonstrate how a blending of fundamental 
and applied research is essential for a scientific study of disease. Cancer will serve as a 
prototype of those diseases open to attack through basic research performed at this 

All participants in this session are exceptionally well qualified for each has con- 
tributed to the basic science of genetics and to problems of practical importance. 
They exemplify the difficulty in drawing a sharp line of demarcation between funda- 
mental and applied scientists. Perhaps the tendency to perform either kind of research 
is dependent upon the genotype of the individual. We heard from Dr. Castle how, as a 
young man, Dr. Little was confronted with a practical problem in tumor transplan- 
tation by Dr. Tyzzer. Investigation of this problem must have impressed Dr. Little 
with the importance of inbred animals in cancer research and this idea led to the 
development of inbred strains of mice that are not only basic research tools used by 
cancer workers but also are rapidly affecting other branches of medical science. 

This morning we shall hear how the use of these strains has aided the progress of 
cancer research. It is reasonably safe to assume we shall learn that the occurrence of 
lung tumors in the inbred mouse is largely dependent upon hereditary factors and this 
tumor can be used for precise determinations of the interplay of hereditary and environ- 
mental factors in carcinogenesis. Mammary gland tumors are more complex for they 
arise through the interaction of hereditary, hormonal and environmental influences. 
The round-table session of yesterday emphasized the importance of hormonal stimu- 
lation in the occurrence of a variety of tumors and mammary gland tumors belong 
in this group of neoplasms. Now we shall hear that, in addition to hereditary and 
hormonal factors, a transmissible agent is also involved in the occurrence of mammary 
gland tumors in mice. The discovery of this agent is one of the outstanding con- 
tributions of this Laboratory to cancer research. 

After hearing yesterday's discussion of carcinogenesis and today's session you will 
note that as we progress in cancer research we become more dependent upon the use 
of inbred animals. Indeed, it now appears that these animals may be essential for 
solution of the basic problems of the difference between normal and malignant cells 
and the factors responsible for the change from the normal to the malignant state. This 
does not imply that such knowledge is essential for the prevention or cure of cancer for 
the history of medicine contains many instances of the cure or prevention of disease 
without knowledge of etiologic factors or the fundamental cellular changes involved. 
The Jackson Memorial Laboratory has supplied one excellent example. Discovery of 
the mammary tumor agent made possible the virtual elimination of breast tumors in 
inbred strains of mice. 

We must remember, however, that, for the purpose of this Symposium, cancer serves 
as an example of how the basic investigations from this Laboratory are applicable to 
a practical and pressing problem in medical science. It is doubtful whether the same 
disease will be used as such when some of you attend the fiftieth anniversary program. 
Regardless of the problems you will face, it is reasonably safe to predict that the 
genetic constitution of the hosts will represent an important part of your discussions. 
For, under the inspiring leadership of Dr. Little, the application of genetic principles 
to biologic problems and the establishment of inbred animals as research tools are 
basic contributions made by this Laboratory. 


Localization of Gene Action in the 
Causation of Lung and Mammary 
Gland Tumors in Mice x 

W. E. Heston, National Cancer Institute, 2 
Bethesda, Md. 

Owing in large measure to research carried out here at the Jackson 
Laboratory it is now almost universally accepted that genes are involved 
in the causation of cancer. The general pattern of inheritance for the 
various types of cancer has been quite clearly defined as multiple factor 
inheritance with alternative expression of the character. Primary prob- 
lems of the future lie in the field of physiologic genetics where attempts 
are made to link the gene to the character by describing the paths through 
which gene action influences the probability that the tumor will occur. 

The first step in this general area is obviously the localization of the 
gene action, the subject of this discussion. Such localization may be in 
respect to anatomy, i.e., in determining in what organs or tissues the gene 
action is occurring; in respect to physiology, i.e., what physiologic path- 
ways are initiated or directed by the gene action; and in a broad sense 
such localization may include the possibility of gene changes at the time of 
the neoplastic change of the cell. Certain approaches to these problems 
will be described and illustrated by projects carried out with respect to 
mammary gland and pulmonary tumors of the mouse. There is no definite 
claim that these approaches are new as labeled in the title of this morning's 
session. So-called new approaches are often old approaches modified to 
fit new situations and answer new questions. The chief concern, however, 
is whether or not the information obtained is new, and from the studies 
referred to herein certain new information has been forthcoming. 

The transplantation of genetically different organs or tissues from two 
inbred strains into a common host, the Fi hybrid of those two strains, 
provides a means for determining whether or not certain gene action 
which controls characters that may develop later in these organs is local- 
ized in the respective organs or tissues. If while in a common host the 
genetically different organs or tissues retain their difference in respect to 
the later occurrence of the character, one could assume that the action of 
the genes controlling this character is being manifest within the trans- 
planted organ or tissue or that some characteristic relative to the later 
occurrence of the character has been established in the organ or tissue and 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 30, 1954. 
2 National Institutes of Health, Public Health Service, U.S. Department of Health, Education, and Welfare. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

776 proceedings: symposium on 25 years of 

has persisted after transplantation. On the other hand, if the difference 
is eliminated and the occurrence of the character is characteristic of the 
host one would assume that the action of the genes is being manifest 
through some general systemic mechanism. Transplantation of the organs 
or tissues at various ages could give information on sequence of events 
in the gene-action path. 

This type of study can be illustrated by an experiment we have carried 
out in regard to pulmonary tumors (1). It had been estimated earlier 
that the susceptible strain A and the resistant strain C57L differed by at 
least four pairs of genes controlling the occurrence of pulmonary tumors 
(2). Then, the question to be answered was: Did these genes have their 
primary action in the tissues of the lung or was their action manifest 
through some general systemic mechanism? Transplants of small por- 
tions of the lungs of adult strain A mice and of adult strain C57L mice 
were made subcutaneously in the axillae of a common host, the (A X C57L) 
Fi hybrid. After allowing time for the transplants to become vascularized 
the host was injected intravenously with a carcinogen 1, 2, 5, 6-dibenz- 
anthracene. Twelve to fifteen months following the injection, the Fi 
animals were necropsied and the transplants were recovered. Serial sec- 
tioning of the transplants revealed that whereas many of the susceptible 
strain A lung transplants had developed typical lung tumors, very few of 
the genetically resistant strain L transplants had done so, although they 
were in a common host. Thus, instead of the genotype of the host govern- 
ing the occurrence of the tumors, their occurrence was controlled by the 
genotype of the donors acting within the transplanted tissue or by having 
fixed a characteristic of the tissue that persisted following transplantation. 

This latter possible explanation was later tested by transplantation of 
lobes of fetal lungs into adult F x hosts. 3 These fetal transplants proved 
to be very interesting in their growth patterns in that, although very 
small when transplanted, many proceeded to grow until in many cases 
they were approximately the size of lobes of the lungs of adults, at which 
time growth ceased. Furthermore, there were many interesting histo- 
logic changes in these transplants which Dr. C. H. Steffee plans to study 
in detail later and which will therefore not be discussed here. The 
general outcome of this experiment, however, was similar to that in which 
adult tissues were used. The occurrence of tumors in the transplants 
was characteristic of the donor strains and not of the Fi host. This fact 
would give further support to the idea that the primary action of these 
genes controlling the occurrence of lung tumors is in the lung tissue. 

More tumors were found in these susceptible fetal transplants than in 
the adult transplants, evidently just because the transplants had grown 
and had provided more tissue in which tumors could arise. Many trans- 
plants had multiple nodules. Furthermore, some of the tumors occurring 
in the fetal transplants were larger than those found in the adult trans- 
plants. A number of the larger tumors were transplanted into the strain 
of origin of the original transplant and all grew progressively. They are 
still being carried after many transplant generations. 

» Unpublished data. 

Journal of the National Cancer Institute 


In this second experiment some of the hosts were injected with the 
carcinogen, whereas others were not. The difference in response of the 
transplants due to the carcinogen was similar to the response in the lungs 
of the donor strains. 

This type of experiment can be and has been adapted to many different 
situations. Shapiro and Kirschbaum (3) have carried out such an ex- 
periment using the susceptible Bagg albino strain and the resistant strain 
DBA. Lung transplants from day-old mice were inoculated into the ears 
of the Fi hybrids, and after receiving the transplants the Fi hybrid hosts 
were injected with urethan. The results of this study indicated that the 
genes controlling the occurrence of lung tumors, and by which these 
strains differed, also had their primary action in the lung tissue. It has 
not been estimated by how many such genes these two strains differ, nor 
is it known which, if any, are the same genes as those by which strains 
A and C57L differ. 

The same technique can, however, be used in respect to specific genes, 
as is being done in our laboratory with the lethal yellow gene that is 
known to increase the occurrence of pulmonary tumors. This experiment 
has not been completed. It would appear that significant results may 
be more difficult to obtain here, owing to the relatively small difference 
effected by only one specific gene. The outcome of this test, however, will 
be of particular interest since in this case we are dealing with a gene known 
to have effects elsewhere in the body, i.e., controlling pigmentation of the 
hair, causing obesity, and in some unknown way having a lethal effect in 
embryos homozygous for the gene. 

Such techniques can be readily applied to other organs. For example, 
Huseby and Bittner (4) showed by transplantation of adrenal glands that 
the genotype of the transplanted adrenal gland governed whether the 
adrenals became carcinomatous or merely hyperplastic following castra- 
tion. In his discussion yesterday, Dr. Woolley (5) described similar 
results using strains CE and DBA. This is particularly interesting in 
this system where physiological interorgan relationships are so prominent. 
Law (6) applied similar techniques in his study of the occurrence of 
leukemia in Fi hybrids bearing thymic transplants from the parent strains. 
He encountered a very interesting situation here in that only the stroma 
of the transplant persisted, whereas the transplanted thymocytes were 
apparently replaced by invasion of those from the host. The resulting 
leukemias were not transplantable back into the donor strain, but were 
transplantable to other of these Fi hybrids. Similar techniques have 
been applied to the problems of mammary gland tumors, and these will 
be described later. Approaches in which tissue culture is substituted for 
the common F x host would be of interest. Strains differ in respect to 
occurrence of subcutaneous sarcomas. It would be interesting to see 
whether or not this genetic difference would persist in cultured fibroblasts 
from the different mouse strains. 

In the problem of mammary gland tumors in mice, attempts at locali- 
zation of gene action have been more toward physiologic pathways through 

Vol. 15, No. 3, December 1954 

778 proceedings: symposium on 25 years op 

which the action becomes manifest. At the present time one can visualize 
three such paths: 1) the control of the propagation and transmission of 
the mammary tumor agent; 2) the control of production of the hormonal 
stimulation; and 3) the control of the response of the mammary tissue 
either to the mammary tumor agent or to the hormonal stimulation. 

Prehn (7) transplanted mammary glands from both resistant strain 
C57BL and susceptible strain BALB/c females into (C57BL X BALB/c) F x 
hybrids. Following the introduction of the mammary tumor agent, 
tumors and hyperplastic nodules arose in the transplanted BALB/c glands 
but not in the transplanted C57BL glands, indicating that there was some 
genetic control over the response of the mammary tissue. But with 
these strains it is not certain whether this response is to the stimulation 
by the agent or to the hormonal stimulation. In an experiment we now 
have in progress, mammary-gland transplants have been made from 
strain C3H with a high incidence of mammary tumors in virgin females, 
and strain A with a low incidence in virgin females, into their (C3H X A)Fi 
hybrids which were kept as virgins. Thus far only one mammary tumor 
has arisen in a transplanted gland and it arose in a gland from strain C3H. 
The fact that the tumor was transplantable back to strain C3H added 
evidence of its C3H origin. Since the genetic difference between these 
strains is in respect to the hormonal stimulation the study should give 
evidence as to whether the genes were controlling the response of the 
gland to the hormonal stimulus or controlling the output of the stimulus. 

Huseby and Bittner (8) approached this problem through transplanta- 
tion of ovaries. They found that (C3H X A)Fi spayed females bearing 
either C3H or F] ovaries had a higher incidence of mammary tumors with 
a lower tumor age than did the same type of F! females bearing A ovaries. 
This would indicate that the action of certain genes occurs within the 
ovaries, or some other organ of the endocrine system, thus controlling the 
hormonal stimulation. It was also indicated, however, that other genes 
may be acting within the cells of the mammary glands in controlling their 
response to the hormonal stimulation, for there were more tumors in F t 
females bearing A ovaries than in A females bearing transplanted A 

Of major interest to us, however, is the relationship between the gene 
or genes and the mammary tumor agent. Definite control of the genotype 
over the propagation and transmission of the agent was demonstrated 
through comparison of backcross females from the original cross of high- 
tumor strain C3H females with low-tumor strain C57BL males (9). 
Females resulting from backcrossing the Fi females to C3H males not 
only had a higher incidence of mammary tumors than did the females 
resulting from backcrossing the F x females to C57BL males, but they also 
transmitted the agent to test females more effectively than did the C57BL 
backcross females. BC3H F! test females that nursed upon the C3H 
backcross females had a higher incidence of tumors than did those that 
nursed upon the C57BL backcross females. Thus, genetic control over 
the agent was indicated. It was suggested that multiple genes were 

Journal of the National Cancer Institute 


involved since variation of the C57BL backcross appeared to be con- 

Subsequently a more extensive analysis of this relationship between 
the genes and the mammary tumor agent has been made. 4 This was the 
study to which Dr. Dunn made reference in her paper. It will be reported 
in detail in the near future. In this study again strain C3H females 
were outcrossed to C57BL males and this outcrossing was followed by a 
series of 7 generations of backcrossing to C57BL males. By this pro- 
cedure a C57BL genetic background was built up, while at the same time 
a continuous line of females through which the agent might be transmitted 
was maintained. Not only was the tumor incidence for each generation 
tabulated, but the females of each generation were tested for their ability 
to transmit the agent to foster-nursed C3Hf test females without the 
agent. From the earlier study it was anticipated that such procedure 
would probably eliminate the agent. Furthermore, Murray and Little 
(10) had reported earlier that the agent was rendered noneffective in the 
eighth generation of backcrossing to C57BL males subsequent to out- 
crossing strain DBA females to C57BL males. By testing each generation 
as was done in the present experiment it was possible to determine just 
how readily the agent could be eliminated. 

By the second backcross generation the incidence of mammary tumors 
had been reduced to 5 percent in the breeding backcross females and zero 
in the virgins. The 5 percent represented one animal, and its tumor 
arose at 21 months of age. These data themselves would have suggested 
that by this generation the agent had been eliminated had not 19 percent 
of the C3Hf test females foster nursed by these second-backcross-genera- 
tion females developed tumors at an average age of 13 months. Further 
inspection of the data revealed that two of the second backcross females 
that did not themselves develop tumors were transmitting the agent to 
their test females. Thus, it could not be considered that the agent had 
been completely eliminated by the second backcross generation. 

In the third backcross generation, however, none of the breeding 
females developed tumors. Only one of the virgins developed a tumor 
and it arose when the animal was 26 months of age. Furthermore, of 
the 93 C3Hf test females foster-nursed by these third backcross females 
only one developed a tumor and it arose when the female was 21 months 
old. One tumor occurred in a later backcross generation and a few oc- 
curred in the C3Hf females used to test the later generations but all of 
these occurred when the females were very old. Thus, it was evident 
that the agent had been eliminated or rendered inactive by the third 
backcross generation. 

The fact that the agent was eliminated by as early as the third back- 
cross generation is highly significant in that it indicates that not many 
genes could be involved in the control of the agent, and even suggests the 
possibility that this relationship may be between but one gene and the 

* Unpublished data. 

Vol. 15, No. 3, December 1954 

780 proceedings: symposium on 25 years op 

agent. This would be in line with the gene-Kappa relationship described 
by Sonneborn and co-workers in Paramecia. If there were delay in 
elimination of the agent in animals without the gene, such as has been 
observed with respect to Kappa, this could explain the failure in the earlier 
study, to observe single gene segregation in the first backcross generation. 
Andervont (11) found that whereas strain C57BL females that received 
the agent could transmit it to susceptible foster-nursed mice, their C57BL 
female progeny could not transmit it. This would suggest one-generation 
delay in the elimination of the agent in mice without the genetic factors 
necessary for its survival. In such a situation the genotype of a female 
could not be judged as well by the occurrence of tumors in her foster- 
nursed test females, as by the occurrence of tumors in the test females that 
nursed upon her female progeny. Examination of the present data from 
this viewpoint gave further suggestion that a single-gene difference might 
be involved. It is hoped that this possibility of but a single-gene relation- 
ship to the agent can be tested further. 

To determine whether or not through alteration in genotype the agent 
had been eliminated or merely rendered inactive, females of the seventh 
backcross generation were outcrossed to strain C3Hf males without the 
agent and this was followed by a series of backcrosses to C3Hf males 
through four generations. By restoring the C3H genotype one could 
expect to activate an agent that had been inactivated by the reverse 
procedure. An interesting assortment of mammary tumors occurred in 
the females of this series but the incidence of tumors was low and did not 
increase as proportion of C3H chromatin background was increased in 
successive generations. Furthermore, the tumors came up when the 
females were very old, so there was no suggestion that the agent had been 
activated. It was, thus, evident that by eliminating the necessary gene, 
or genes, the agent likewise had been eliminated and not caused to mutate 
to a latent or inactive state. 

At the time this experiment was begun the possibility that the agent 
might have arisen through mutation of some normal cell component was 
a rather popular consideration, despite the fact that this was based in 
large measure upon speculative thinking. To test this possibility recip- 
rocal series of backcrosses were produced. Strain C57BL females were 
outcrossed to C3H males and this outcross was followed by seven 
generations of backcrossing to C3H males. Because of the possible 
introduction of the agent by the C3H males a second series was produced 
using C3Hf males. 

Although a few tumors occurred in the C3Hf backcross series, in no 
generation was the incidence high but remained low despite the increase 
in proportion of C3H chromatin built up in the successive backcross 
generations. Furthermore, in all cases the tumors arose in females of 
advanced age indicating that the tumors had resulted from causes other 
than the milk agent. In the C3H backcross series, however, more mam- 
mary tumors arose, the incidence increasing and average tumor age 
decreasing in successive generations until, with the exception of one line, 

Journal of the National Cancer Institute 


the seventh generation was comparable to strain C3H in both incidence 
and tumor age. This was the result of introduction of the agent by the 
males. In the pedigree chart each point at which the agent was intro- 
duced was evident, for once introduced the female progeny then developed 
tumors as females of any high-tumor line would develop them. 

Some interesting observations were made in the classification of the 
mammary tumors that arose in these series. These were classified by 
Dr. Thelma Dunn according to the classification she has been using (12) . 
The greater proportion of tumors were adenocarcinomas of Type A and 
Type B. For the most part these were the types that arose in the presence 
of the agent and in the younger females. As the agent was eliminated 
in the C57BL backcross series the proportions of these two types were 
diminished; the proportion of the adenocarcinomas of Type C, of the 
adenoacanthomas, of the molluscoid variation of the adenoacanthoma, 
and of the carcinosarcomas increased. These latter four unusual types 
were particularly numerous in the series resulting from outcrossing the 
seventh generation C57BL backcross females to C3Hf and backcrossing 
to C3Hf males. In the C3Hf backcross series the unusual types were 
scattered through the series along with a few adenocarcinomas of Type A 
and Type B. In the C3H backcross series the proportion of adenocar- 
cinomas of Type A and Type B was low in the early generations but 
increased as the mammary tumor agent was introduced and the tumor 
age was decreased. An extremely interesting study is thus opened in 
respect to relationship between the various causative factors and the 
resultant histologic type of tumor. It is assumed that genes influencing 
occurrence of mammary tumors through control over the milk agent have 
little or nothing to do with the occurrence of these unusual types that 
develop in the absence of the agent. It would be of interest to know 
whether action of other specific genes might be associated specifically 
with some of these unusual histologic types of tumors. 

The question of whether or not the malignant change is the direct result 
of a change in a gene of the cell that becomes malignant, i.e., a somatic 
mutation, has been studied through many approaches. These have all 
been indirect approaches, however, as they will continue to be until some 
means of observing segregation of genes in somatic cells is devised. One 
of the approaches in which we have been particularly interested has been 
through analysis of dose response. The shape of the response curve 
should give some indication of events at the time of the malignant change. 
If the curve were exponential one could assume that more than one in- 
dependent event were necessary for the change, whereas a linear response 
would indicate the result of only one event. If the malignant change 
were due to a recessive mutation, two mutations at the locus concerned 
would be necessary and thus the curve would be exponential. If the 
malignant change were due to a dominant mutation only one mutation 
at the locus concerned would be necessary, and one would expect the 
dose-response relationship to be linear. 

Charles and Luce-Clausen (13) analyzed the dose-response data ob- 

Vol. 15, No. 3, December 1954 

782 proceedings: symposium on 25 years of 

tained by Morton, Luce-Clausen, and Mahoney in inducing papillomas 
in strain DBA mice by successive paintings of the skin with methyl- 
cholanthrene. When the square root of the average number of papillomas 
was plotted against time, in this case a measure of dose, a straigho line 
was observed, and it was pointed out that this was what could be expected 
if the papillomas arose from a recessive mutation in the treated cells of 
the skin, thus requiring two independent events at the locus concerned. 

Induced pulmonary tumors offer excellent material for such a study. 
The multiple independent tumors arising following the injection of a 
carcinogen provide an excellent quantitative measure of response. In an 
early experiment (lJf) we injected groups of strain A mice with graded 
doses of 1,2,5,6-dibenzanthracene ranging from 0.1 mg. to 0.5 mg. in an 
aqueous colloidal dispersion. Six months after the injection the animals 
were killed and the number of tumor nodules appearing on the surface of 
the lungs of each animal was recorded. When average number of nodules 
in each group was plotted against dosage, a linear relationship was ob- 
served. Extension of this line did not, however, pass through the average 
number in untreated animals, but intersected the response scale at a point 
below zero. Investigation of response between zero and 0.1 mg. dose 
was therefore indicated. 

This area of the dose range has now been investigated. 5 Groups of 
strain A mice were injected with 0.01 mg., 0.02 mg., 0.04 mg., 0.06 mg., 
0.08 mg., 0.1 mg., and 0.5 mg. of 1,2,5,6-dibenzanthracene respectively. It 
was necessary, however, to use a new aqueous colloidal dispersion of the 
dibenzanthracene for these injections. Because of variation in ages of 
the animals, two series were injected at an interval of approximately 2 
months. When the mice of the first series were killed at 6 months after 
the injection, the dose response relationship was again found to be linear 
within these dose limits. Mice of the second series have now been killed 
after the same time interval and again a linear relationship was observed 
between response and dose. Combination of the two series even improved 
the approximation to a straight line. 

In the second experiment, as in the first, the average number of nodules 
at zero dose deviated significantly above the extension of the dose- 
response curve. This deviation in both experiments was probably due 
to some of the carcinogen of each dose passing through the lungs. It 
would seem logical to assume that some of the smaller crystals would not 
be lodged in the capillaries of the lung. The deviation at the zero dose in 
the second experiment, however, was not as great as that in the first. 
This was probably because two dispersions were used and the particles 
of the second dispersion were probably larger than were those of the first 
dispersion. With the colloidal mill it is difficult to get two dispersions 
with the same particle size. Variation in particle size was also indicated 
by the fact that a considerably greater response was noted at the 0.1 mg. 
and 0.5 mg. doses in the second experiment than in the first. Shim kin 
and Lorenz (15) have shown that a greater response does result when the 
size of the particles is larger. 

» Unpublished data. 

Journal of the National Cancer Institute 


These observations include responses throughout the practical dose 
range. Higher doses would give so many nodules that counting could not 
be accurate. Furthermore, the number of nodules is limited by the 
amount of lung tissue, and thus, extension of the line would be expected 
to curve downward. Whether or not a linear relationship would be 
observed for other time intervals could be ascertained only by further 
experimentation. It is interesting that within this practical dosage range 
and at this time interval a linear dose-response relationship was observed. 
This is in line with a mechanism involving a single happening in the cell 
transforming to the malignant cell. If this happening were a change in 
a gene it would thus be a dominant mutation. 


(1) Heston, W. E., and Dunn, T. B.: Tumor development in susceptible strain A 

and resistant strain L lung transplants in LA Fi hosts. J. Nat. Cancer Inst. 
11: 1057-1071, 1951. 

(2) Heston, W. E.: Genetic analysis of susceptibility to induced pulmonary tumors 

in mice. J. Nat. Cancer Inst. 3: 69-78, 1942. 
(8) Shapiro, J. R., and Kirschbaum, A.: Intrinsic tissue response to induction of 
pulmonary tumors. Cancer Res. 11: 644-647, 1951. 

(4) Husebt, R. A., and Bittner, J. J.: Differences in adrenal responsiveness to post- 

castrational alteration as evidenced by transplanted adrenal tissue. Cancer 
Res. 11: 954-961, 1951. 

(5) Woolley, G. W.: Carcinogenesis in the adrenal. J. Nat. Cancer Inst. 15: 

717-719, 1954. 

(6) Law, L. W.: Increase in incidence of leukemia in hybrid mice bearing thymic 

transplants from a high leukemic strain. J. Nat. Cancer Inst. 12: 789-806, 

(7) Prehn, R. T\: Tumors and hyperplastic nodules in transplanted mammary 

glands. J. Nat. Cancer Inst. 13: 859-872, 1953. 

(8) Huseby, R. A., and Bittner, J. J.: Studies on the inherited hormonal influence. 

Acta Unio Internat. Contra Cancrum 6: 197-205, 1948. 

(9) Heston, W. E., Deringer, M. K., and Andervont, H. B.: Gene-milk agent 

relationship in mammary-tumor development. J. Nat. Cancer Inst. 5: 
289-307, 1945. 

(10) Murray, W. S., and Little, C. C: Chromosomal and extrachromosomal 

influence in relation to the incidence of mammary tumors in mice. Am. J. 
Cancer 37: 536-552, 1939. 

(11) Andervont, H. B.: Fate of the C3H milk influence in mice of strains C and C57 

black. J. Nat. Cancer Inst. 5: 383-390, 1945. 

(12) Dunn, T. B. : Morphology of mammary tumors in mice. In The Physiopathology 

of Cancer, (Hamburger, F., and Fishman, W. H, eds.), New York, Hoeber-Harper, 
1953, Ch. 8. 

(13) Charles, D. R., and Luce-Clausen, E. M.: The kinetics of papilloma formation 

in benzpyrene-treated mice. Cancer Res. 2: 261-263, 1942. 

(14) Heston, W. E., and Schneiderman, M. A.: Analysis of dose-reponse in relation 

to mechanism of pulmonary tumor induction in mice. Science 117: 109-111, 

(15) Shimkin, M. B., and Lorenz, E.: Factors influencing the induction of pulmonary 

tumors in strain A mice by carcinogenic hydrocarbons. J. Nat. Cancer Inst. 
2:499-510, 1942. 

Vol. 15, No. 3, December 1954 


Dr. Leon Dmochowski, 1 College of Physicians and Surgeons, Columbia University, 

New York, N. Y. 

It is an honor and pleasure for me to participate in this meeting of the Roscoe B. 
Jackson Memorial Laboratory and I should like to express my sincere thanks to Dr. 
C. C. Little and Dr. Elizabeth S. Russell for their kind invitation, and to all who made 
it possible for me to be present at this memorable meeting. 

I would like to congratulate Dr. W. E. Heston on his penetrating and stimulating 
analysis of the localization of gene faction in the origin of lung and mammary cancer 
of mice. There is indeed little, if anything, I can add to what he has presented to us. 
Of necessity, dictated by lack of personal experience, I shall limit myself in what I 
will say to observations on mammary cancer. In the few observations I propose to 
discuss, I will attempt to show the possible part played by genes in the origin of mam- 
mary cancer, and will be grateful to hear Dr. Heston's and others' comments on the 
possible action of genes in the development of breast cancer recorded in these 

As you have heard, one of the paths along which the action of genes may manifest 
itself, is the control of the propagation and transmission of the mammary-tumor 
inciter. You have also heard that there may even be a single-gene agent relationship, 
in spite of the variation in ability to transmit the agent in the backcross generations. 
Here, I should like to mention the following observation (1-4)- After the mating of 
low-cancer-strain females with high-cancer-strain males, combined with brother X 
sister matings of their hybrid progeny and subjecting this progeny to increased hor- 
monal stimulation, a variable number of mammary tumors has been observed [by 
ourselves and also by others (9-11)] to develop in successive generations of this 
hybrid progeny. Biological tests of these tumors, which we have carried out, revealed 
the agent in some very young as well as in very old tumors, while in other young and 
old tumors the same tests failed to demonstrate the agent. These tests, showed that 
under the same experimental conditions among hybrids of the same derivation some 
developed tumors which either harbored the agent or which failed to reveal the agent, 
while other hybrids even their litter mates died without tumors. Some of the progeny 
of tumorous Fi hybrids failed to develop breast cancer, while other progeny of the 
same hybrids developed cancer in which the agent was either demonstrated or could 
not be shown. Some of the progeny of tumor-free Fi hybrids developed breast tumors 
that either harbored the agent or did not reveal the agent. Examination of family 
charts of the progeny of both tumorous and nontumorous females reveals, perhaps, 
the variation in the susceptibility to tumor development on one hand, and on the 
other, the variation in propagation and transmission of the agent in hybrid progeny 
of identical origin. It also appears that the different behavior of tumors in the 
bioassays may be an expression of the gene control over the relationship of the agent 
to the substrate, that is, the mammary-gland cell as a whole, or the known and as yet 
perhaps unknown components of the cell. In other words, assuming the presence of 
similar amounts of the agent in the tumors tested, the difference in the bioassay be- 
havior of the tumors may be based on variation in the integration of the agent with 
cellular constituents, an integration controlled by genetic factors either directly or 
indirectly through their control of the response of the mammary-gland cell to hormonal 
stimulation or both. 

A somewhat more puzzling observation is the development of breast cancer in the 
hybrid progeny of two agent-free, low-cancer strains which had been maintained by 
brother X sister matings and subjected to increased hormonal stimulation. These 

i On leave of absence from Leeds University, Leeds, England. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

316263—54 35 

786 proceedings: symposium on 25 years of 

tumors have been found to contain the agent {12). Although the incidence of tumors 
observed was much lower than that in the hybrid progeny of low- and high-breast- 
cancer strains, the question that still remains to be answered is, how does the agent 
originate in these tumors? Is it possible that crossing of genes from two different 
strains known to be agent-free results in a substrate more sensitive to hormonal 
stimulation than the genetic constitution of either of the parental strains? Can, 
therefore, the combination of a suitable genetic background with appropriate hormonal 
environment result in the appearance of the agent? I hesitate to answer this in the 
affirmative. Even so, I do not know if this would not be better than the acceptance 
of a universal agent, latent or otherwise, to be revealed by some suitable genetic 
combination and suitable internal and/or external environments. 

To continue on this puzzling trend, I would like to mention the observation of an 
increased incidence of breast cancer in a subline of strain C57BL mice and the eventual 
demonstration of the agent in tumors of mice of later generations of this subline (IS). 
This may be seen in a simplified version on a chart, part of which is known to some 
of you. When three years ago I discussed this observation with Dr. Heston, he urged 
me to study the behavior of virgin females of this subline. We have now found that 
these females do not develop breast cancer, thus revealing the importance of the 
hormonal stimulation in the origin of breast cancer in this subline. However, this 
still does not answer the problem of the origin of the agent unless we assume either 
its latency in the preceding generations — where again we may stumble on the agent's 
ubiquity or that a mutation of gene or genes had taken place. This latter possibility 
could not be verified by transplantation as the tumors grew in mice of all other sub- 
lines of strain C57BL, which, of course, does not exclude the possibility of a genetic 
change. Finally, we may think of endogenous origin based on a change in a normal 
cell component, either independently or following a genetic change directly or indirectly 
through hormonal factors. What the answer is I don't know, but it may possibly 
lend itself to a discussion on the probable part played by genes in the appearance of 
the agent. 

There is one more observation made recently by us in the study of the appearance 
of breast-tumor cells in thin sections seen in the electron microscope (14). Following 
the observation of characteristic bodies in the cytoplasm of tumors from high-cancer 
strains, obtained at first from our own and then also from several widely separated 
laboratories (fig. 1) , tumors from low-cancer strains have been studied. In the absence 
of these bodies in breast tumors of a low-cancer strain in our own laboratory, we 
obtained a number of tumors from an agent-free strain, the so-called C3Hf, from Dr. 
Heston and also tumors from agent-free strain C3H mice born from transplanted ova 
and supplied by Dr. Fekete. Similar bodies to those present in high-cancer-strain 
tumors, were found in 4 of 8 tumors obtained from Dr. Fekete and also in 7 of 14 tumors 
supplied by Dr. Heston. This seemed to be disappointing as far as any conclusions 
about the significance of these bodies were concerned. However, each of these tumors 
studied in the electron microscope was also tested biologically. Three of the 4 tumors 
from Dr. Fekete have already been found to harbor the agent, and 2 of the 7 containing 
the characteristic bodies and obtained from Dr. Heston have so far been found to be 
active in the bioassay. It is too early as yet to judge the bioassays of the remaining 
tumors. As to the nature of these bodies, the first question which we must ask our- 
selves is, are they normal cell constituents? It appears to us on the basis of all the 
available evidence that they are not. The next problem confronting us is the possible 
connection between the presence of these bodies and the presence of the agent. If 
there is a connection, are they the agent itself or are they an expression of a functional 
state of the cell which harbors the tumor-inducing activity? Admittedly, these 
particles do look like viruses reported by others in various cells. Before fulfilling 
Koch's postulates and combining them with suitable electron-microscope studies it 
is impossible to answer the question whether these bodies are the agent itself. There 
is, however, the problem of the undoubted tumor-inducing activity of some of the tu- 
mors as shown in the biological tests. This may well be the result of a genetic change, 

Journal of the National Cancer Institute 


which may lead to a functional state of the cell which in turn is manifested biologically 
by the presence of tumor-inducing activity in these cells. The connection between 
the tumor-inducing activity and these bodies — that is, between the two properties 
of a cell, the morphological and biological — although as yet not finally established by 
us, appears to be at least interesting to contemplate as a possible basis for correlation 
of the gene action, in the morphological and physiological sense. 


(1) Dmochowski, L.: In 27th Ann. Rep. Brit. Emp. Cancer Campaign 27: 162, 1949. 
In 27th Ann. Rep. Brit. Emp. Cancer Campaign 28: 169-170, 1950. 
In 27th Ann. Rep. Brit. Emp. Cancer Campaign 29: 150-151, 1951. 
A study of the development of mammary tumours in hybrid mice. Brit. 



J. Cancer 7: 73-119, 1953. 

(5) Andervont, H. B., and Dunn, T. B.: Mammary tumors in mice presumably 

free of the mammary-tumor agent. J. Nat. Cancer Inst. 8: 227-233, 1948. 

(6) : Efforts to detect a mammary- tumor agent in strain C mice. J. Nat. 

Cancer Inst. 8: 235-240, 1948. 

(7) : Further studies on the relation of the mammary-tumor agent to mam- 
mary tumors of hybrid mice. J. Nat. Cancer Inst. 9: 89-104, 1949. 

(8) : Attempt to detect a mammary-tumor agent in strain C mice by X- 

radiation. J. Nat. Cancer Inst. 10: 1157-1190, 1950. 

(9) Bittner, J. J.: Transfer of the agent for mammary cancer in mice by the male. 

Cancer Res. 12: 387-398, 1952. 

(10) Foulds, L.: Mammary tumors in hybrid mice: the presence and transmission 

of the mammary tumor agent. Brit. J. Cancer 3: 230-239, 1949. 

(11) MtiHLBOCK, O. : Studies on the transmission of the mouse mammary tumor agent 

by the male parent. J. Nat. Cancer Inst. 12: 819-837, 1952. 

(12) Dmochowski, L.: Unpublished data. 

(13) : Studies on spontaneous mammary tumors in a low breast-cancer strain 

of mice. (Abstract.) Proc. Am. Assn. Cancer Res. 1: 11-12, 1954. 

(14) Dmochowski, L., Haagensen, C. D., and Moore, D. H.: Electron microscope 

studies of thin sections of normal and malignant mammary cells of some high 
and low cancer-strain mice. (Abstract.) Proc. Am. Assn. Cancer Res. 
1: 12, 1954. 

Vol. 15, No. 3, December 1954 


proceedings: symposium 

Plate 47 

Figure 1. — Section of a high-cancer-strain mammary tumor. Characteristic bodies 
present in a duct and showing an internal structure composed of an inner dense region 
surrounded by an outer paler zone, which in turn is surrounded by a dense outer 
region. X 67,000 






a * 

• ■ • I 




.. . 




Figure 1 

The Use of Fi Hybrid and Backcross 
Generations to Reveal New and/or 

Uncommon Tumor Types h 2 

Margaret M. Dickie, 3 Roscoe B. Jackson 
Memorial Laboratory, Bar Harbor, 

In celebrating 25 years of progress in mammalian genetics and cancer 
one must pay tribute to the laboratory and to its founders, who have 
played a major role in the development of the inbred strains of mice 
whose biological uniformity have contributed so much to this progress. 
The search for biological material to meet the needs of present-day re- 
search is never ending. Perhaps one of our greatest assets to date is this 
inbred-strain mouse. Here is an example of biological material with dif- 
fering potentials that have been so fixed by inbreeding that they can be 
reproduced in unlimited quantities, each one as identical to the next as it 
is possible for biological material to be. 

Take for example strain C3H: here is a strain that for some 30 to 40 
generations has produced 98 percent incidence of mammary tumors. 
Strain A has produced a very high incidence of lung tumors over many 
generations. Why these incidences have not been 100 percent still re- 
mains a genetic problem, in part at least. 

To go on with a reverse example: strain C57BL is resistant to mammary- 
tumor development and to the milk agent, a perfect control for strain 

Our list can be increased by including such things as strains which 
differ in the numbers of presacral vertebrae (1), and in resistance or sus- 
ceptibility to specific viruses and bacteria. 

So the values of our inbred strains need no further elaboration. How 
do we develop these inbred strains? Haphazard mating of mice would be 
to no avail but a logical procedure of crossing one inbred strain with 
another inbred strain, of which the genetic background is known, is an 
organized method of arriving at the creation of a new strain. 

1 Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, 
Maine, June 30, 1954. 

* This investigation has been supported by grants to the Roscoe B. Jackson Memorial Laboratory from the 
American Cancer Society upon recommendation of the Committee on Growth of the National Research Council 
and by research grant C-362 from the National Cancer Institute of the National Institutes of Health, U. S. Public 
Health Service. 

3 Grateful acknowledgment is made of the valuable assistance of Mrs. P. W. Lane in collecting the data for this 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

792 proceedings: symposium on 25 years of 

Hybridization is known to produce conditions unknown in either parent 
or conditions more extreme than those found in either parent (2) . Utilizing 
these facts it may then be possible to develop strains of mice showing 
perhaps a new tumor type or pathologic condition. In addition, the Fi 
animal may show dominance or lack of dominance of a condition already 
apparent in the parental strains. This is the same technique used to 
establish proof that new characters are mutations and to provide a clue 
to the manner of their inheritance. 

Bearing these facts in mind, the first step that the geneticist must take 
is to create the Fx hybrid. Because most of these phenomena described 
above do not occur in 100 percent of the animals, the possibility that 
cxtrachromosomal factors like maternal influence, as well as multiple 
genie effects, can play a part makes it necessary to create reciprocal 
hybrid animals. This in effect tests the maternal influence upon the 
occurrence of any phenomena. An excellent example of this was the 
demonstration of the presence of the milk agent using reciprocal hybrid 
mice (3). 

If the geneticist, once started on such a program, records the results 
of the Fi hybrid animals and wishes to establish the inheritance more 
clearly and ascertain the genes involved in the occurrence of the response, 
he must proceed one step further and, for most efficient utilization of the 
mice, backcross the Fi hybrids to the parental strains. Again reciprocal 
backcrosses may be used for the same reasons reciprocal F x hybrids were 
created. If a single-gene difference were responsible for the genetics of 
the character in question, the result of this backcross would be its linkage 
data with the genes against which it was tested. In most instances when 
inbred strains are being tested for a response however, the genetic analysis 
is far more complicated since several genes are probably concerned with 
many physiological responses. By application of certain known formulae 
to the data an estimate of the minimum number of genes involved in 
producing any given response may be calculated. The need for tagging 
such "in visible gene action" is apparent so that linkage of physiological 
responses with phenotypic characteristics is very desirable. The linkage 
of two genes, fused tail and histocompatibility-2, is an excellent demon- 
stration of a phenotypic tag for a physiologic characteristic (4). 

The values of the hybrid animal have long been recognized by the 
geneticist and therefore the use of hybrids is not really a new approach to 
the study of cancer problems. For example, Dr. Myron Gordon (5) has 
demonstrated this with his hybrid fish and melanoma tumor production. 
Both the hybrids and the backcross mice should be exploited more fully 
not only for the information they can give on expected responses but for 
the information they can furnish as to conditions occurring de novo, 
which may be the basis for creating new inbred strains if the condition is 
not a result of hybridization alone and disappears in the next generation. 

This last statement is based on the results of an investigation begun 
several years ago by Drs. Fekete, Woolley and Little (6). Although the 
original experiment was designed for an entirely different purpose, as often 

Journal of the National Cancer Institute 


happens in this business of research, the results showed that inbred strains 
varied in their response to neo-natal gonadectomy. It is this particular 
response and its ramifications that will be used to demonstrate the truth 
of the statements just presented. 

Strain DBA/2Wy, following gonadectomy, developed nodular hyper- 
plasia of the adrenal cortex and this was accompanied by feminization of 
the accessory reproductive organs (6, 7). Strain CE/Wy, however, pro- 
duced adrenal-cortical carcinomas and concomitant masculinization and/or 
feminization of the accessory reproductive organs following gonadectomy 
(8-10). Hybrids of these two inbred strains always developed adrenal- 
cortical carcinomas after neo-natal gonadectomy, thus demonstrating 
that the carcinoma response predominated over the hyperplasia response 
in these animals (11). The time of occurrence of these changes seemed to 
be dependent upon extrachromosomal factors ; offspring of the CE females 
produced carcinomas earlier than those with DBA mothers. Backcrosses 
to both parental strains have been made but results for gonadectomized 
animals are not yet available. 

Strain C3H/Di, following gonadectomy responded in a manner similar 
to that of strain DBA, i.e., nodular hyperplasia of the adrenal and feminiz- 
ing changes (12). Strain A/Wy showed little change in the adrenal and 
the accessory reproductive organs remained permanently immature after 
gonadectomy (12). Hybrids of these strains responded to gonadectomy 
by production of adrenal-cortical carcinomas (IS). In this combination 
the carcinoma response may have been the result of hybridization per se. 
Backcrosses to these parental strains have also been made but data on the 
gonadectomized mice are too incomplete to have any report at this time. 

Using strain DBA in a cross with another stain, DE/Wy (which although 
developed from strain CE responded to gonadectomy like strain A, i.e., 
no changes in the adrenal or accessory reproductive organs) another hybrid 
was created and responded to gonadectomy by producing not only adrenal- 
cortical carcinomas but also by consistently producing basophil adenomas 
of the pituitary. These basophil adenomas are physiologically active, 
i.e., diagnosis for their presence is the abundance of secretion and alveolar 
development of the mammary gland (14). The adrenal-cortical carci- 
nomas stimulate growth of the mammary glands, but they alone never 
produce the extreme development found in the mice carrying basophil 
tumors as well. In this instance, then, both abnormalities appear to be 
the result of hybridization since neither response has been observed in 
either parent strain. 

Thus the three different sets of hybrids have produced similar responses 
from different backgrounds and for different reasons (straight inheritance 
in one instance and hybridization in the other two cases). Factors in 
addition to the gonadectomy response have been found in the hybrids 
and have caused this investigation to take on an even wider scope. 

One hundred percent of the intact or control hybrid DBA X CE virgin 
females developed a pathologic condition of the uteri that resembled 
eventually the "Swiss cheese" endometrium found frequently in the clinic 

Vol. 15, No. 3, December 1954 

794 proceedings: symposium on 25 years of 

(15). This condition began with cystic glandular hyperplasia that ap- 
peared about 8 months of age and was followed by adenomyosis and 
adenomatous hyperplasia. The same condition of the uterus was apparent 
in the gonadectomized females and occurred at about the same age as in 
the intact animals. Studies carried out by Atkinson and others showed 
that in the intact mice the ovary and not the adrenal was responsible for 
this hyperestrogenic or "hyperovarian" syndrome (16). Recent studies 
have shown that development of this condition may be prevented by 
breeding these mice (17). Under such conditions the uteri have remained 

Investigation of these particular Fi animals is continuing in an attempt 
to elucidate the physiological factors involved in the development of this 
syndrome. Its pattern of occurrence is being observed in the backcross 
animals so that the inheritance of this condition may possibly be clarified. 
Data tabulated so far, showed that in the backcrosses of the DBA X CE 
hybrids to strain CE, only 19 of 114 females or 16.66 percent showed this 
condition. Since only two color types were involved in this backcross, 
data were tabulated according to this factor and in both types the occur- 
rence was 16.66 percent. Comparing the data of the offspring of the inbred 
strain mother with that of the Fi mother showed that the incidence was 
very slightly higher in offspring of Fi mothers. In the backcrosses of 
these DBA-CE hybrids to strain DBA, of the 205 females thus far ex- 
amined only 39 or 19.02 percent have been classified as hyperestrogenic. 
Data are too incomplete to be analyzed according to color type but 
occurrence was much lower when the DBA was the mother (12.28%) than 
when the Fi was the mother (21 .62%). This indicates that inheritance is 
a factor in the occurrence of this condition since it has been transmitted 
beyond the hybrid generation. Incidence in both sets of backcrosses 
indicates that maternal factors also play a role in the manifestatiou of this 

The pituitary basophil adenomas, which occurred consistently in the 
DE-DBA gonadectomized hybrids, occurred in about 60 percent of the 
DBA-CE hybrids, and occasionally in other hybrid crosses that were 
studied (14). This phenomenon, too, appears to be a new condition 
hitherto unknown in any of the parent strains and is thus a product of 
hybridization. Very preliminary data collected on DE and DBA back- 
cross gonadectomized mice indicated that basophil adenomas occurred 
in approximately 30 percent of the DBA backcross animals and in approxi- 
mately 29 percent of the DE backcross mice. This too shows that inherit- 
ance must be an important factor in its occurrence since the condition has 
appeared in another generation. No genetic analysis is yet possible and 
no other tabulation of the data has been made. 

Incidental information was collected on these hybrids concerning the 
occurrence of mammary tumor, lung tumor or appearance of any other 
tumor types. One other hybrid group, studied for an entirely different 
reason, is included in the list given in table 1 . 

Journal of the National Cancer Institute 

Table 1. — Types oj tumors found in F\ reciprocal hybrid mice 



32 mice 
2 lung tumors 
6 lymphoid tumors 


16 mice 

14 lung tumors (87.50%) 

7 mammary tumors (100%) 

1 hepatoma 


29 mice 

5 lung tumors 

6 mammary tumors (42.85%) 
1 papilloma 

1 interstitial-cell tumor testis 


20 mice 

11 lung tumors (55.00%) 

2 hepatomas 

2 lymphoid tumors 


34 mice 

14 lung tumors (41.17%) 
2 adenomas Harderian gland 

2 epidermoid carcinomas 

3 fibrosarcomas 

1 lymphatic leukemia 

2 papillomas 

1 myoepithelioma salivary 

3 hepatomas 

1 giant-cell sarcoma 


111 mice 

7 lung tumors 
3 ovarian tumors 
6 uterine sarcomas 

1 hepatoma 

2 sarcomas 

The occurrence of hyperestrinism in the CE and DBA backcrosses has 
been discussed. Ovarian- tumor incidence in these same backcross series 
gives an excellent example of a condition appearing to a greater extent 
in the backcross generation than in either of the parent strains or in the 
Fi hybrids. Strain CE females had the greatest number of ovarian 
tumors found in any strain. After 20 months of age the incidence was 
34 percent. Such tumors did not occur at all in strain DBA mice. The 
incidence was 33.33 percent in DBA-CE hybrid females and 19.04 percent 
in CE-DBA hybrid females. Totaling all females of the CE backcrosses 
the incidence was 52.00 percent after 20 months of age, an enhanced 
incidence over the 34 percent available from strain CE. When examined 
according to color type, the data showed that 58.62 percent of the extreme 
dilute mice and 42.85 percent of the black agouti mice in this age group 
had ovarian tumors. When data were tabulated according to parental 
type 44.00 percent of the offspring of CE mothers had these tumors and 
60.00 percent of the offspring of the Fi mothers had tumors. In the back- 
crosses to DBA only three ovarian tumors were observed. All three 
tumors were papuliferous cystadenomas. A diagram of the occurrence 
of ovarian tumors in the three generations is presented in text-figure 1 . 

Not only did the numbers of ovarian tumors show an increase over that 
observed in the hybrid and in the parent strain CE, but the tumors in 
many instances appeared to be atypical granulosa-cell tumors and tubular 
adenomas. 4 Histologic study of these tumors is not complete. 

The adrenal glands of these various types of intact backcross mice 
have been examined and present some interesting possibilities for further 
investigation. In some groups 25 percent, or more, intact mice had 
pathologic changes in the adrenal glands. Their incidence was highest 
in the backcrosses to strain CE. These have not all been classified as 

* Diagnoses courtesy of Dr. Hummel and Dr. Murphy. 
Vol. 15, No. 3, December 1954 




279 9 tu. 



CE9 i DBAC? 
219 4 tu. 

DBA9 x CEtf 
95 3 tu. 



959 tu. 



CE9 x (CE-DBAjd 1 

139 6 tu. 

46. 19* 

7e« 3 tu. 42.85% 
6AB 3 tu. 50.00* 

CE9 x (DBA-CE)cf 

1^ 5 tu. 


7c e 4 tu. 57.14% 
5AB 1 tu. 20.00* 

(CE-DBA)9 x CEO" 

139 7 tu. 


7c e 3 tu. 42.85% 
6AB 4 tu. 66.66* 








8c e 7 tu. 



4AB 1 tu. 




DBA9 x Fjtf 

59 1 tu. 


F X 9 x DBJtf 
219 2 tu. 

26 9 3 tu. 

50 9 *26 tu. 

(CE 299 17 tu. 58.62% I AS 219 9 tu. 42.85%) 

Text-figure 1. — Ovarian tumor occurrence after 20 months o) age. 

yet and their physiological action, if any, is not clearly understood. 
Many of the adrenals exhibited nodular lesions that were composed of the 
small subcapsular cells which extended over a large portion of the cortex. 
In some of these nodules, areas of clear hyperplastic cells were seen and 
such areas closely resembled the nodular hyperplasia seen in gonadecto- 
mized DBA mice. Another adrenal tumor of quite a different type was 
found and has been provisionally diagnosed as a differentiated adrenal- 
cortical carcinoma. Medullary disturbances were also found. Medul- 
lary nodules arose but did not expand greatly in size. One pheochromocy- 
toma was also observed. 5 

Other rare tumors were found in the backcross animals. One thyroid 
adenocarcinoma was found in a CE-DBA backcross animal and 3 
ganglioneurofibromas of the optic tract were discovered in DE-DBA 
backcross mice. 5 

A list of the various types of tumors in these different backcross groups 
serves to illustrate again the value not only of hydridization but of the 
backcross animals (table 2). 

The value of the inbred strain, the hybrid and the backcross mouse in 
their role of revealing new conditions and providing information on the 
inheritance of conditions already described|is thus apparent. 

Using this technique of hybrid and backcross generations in an experi- 
ment originally designed to demonstrate the inheritance of the gonadec- 
tomy response, far-reaching results have occurred and are thus sum- 
marized: 1) New conditions have occurred in hybrids which show evidence 
of their inheritance by their appearance in the backcross generations. 
They are a) hyperestrinism in 100 percent of intact hybrid DBA-CE 
virgin females (an excellent experimental tool for the uterine pathologist) 
and b) basophil adenomas of the pituitary in 100 percent of gonadecto- 

» Diagnoses courtesy of Dr. Edwin Murphy. 

Journal of the National Cancer Institute 


Table 2. — Types of tumors found in reciprocal backcross series of mice 

Backcross to CE 




epidermoid car- 
lung tumors 
uterine fibrosar- 

Backcross to A 





lung tumors 
mammary tu- 
lymphoid tu- 


1 cervical sarcoma 

2 interstitial-cell 

tumors, testis. 
26 lymphoid tu- 




Backcross to DE 

346 mice 
40 lung tumors 
2 lymphoid tumors 
1 hepatoma 
8 uterine fibrosar- 
1 cervical sarcoma 

1 hemangioma 

2 chromophobe ad- 

enomas pitui- 

Backcross to C3H 
(with A) 








lung tumors 
mammary tu- 
lymphoid tumor 

Backcross to DBA 





lung tumors 
mammary tu- 
uterine fibrosar- 

Backcross to C3H 

(with C) 

271 females 
67 lung tumors 
12 ovarian tumors 
8 cervical sarco- 
8 uterine fibrosar- 






cervical sarcomas 
lymphoid tumors 
thyroid adeno- 


10 hepatomas 

19 lymphoid tu- 

2 osteogenic sar- 


3 adrenal-cortical 



Backcross to DBA 
(with DE) 









lung tumors 

mammary tu- 


uterine fibrosar- 

cervical sarcomas 

lymphoid tumors 



mized DE-DBA mice (another experimental tool for endocrine research) . 
2) A 20-percent increase in ovarian tumor incidence over that available 
in inbred strains has occurred in certain backcrosses. 3) Several rare 
tumor types have been observed in some of the backcross mice, and 4) 
adrenal-cortical abnormalities have been noted in intact backcross mice. 

Vol. 15, No. 3, December 1954 



Needless to say, the occurrence of these particular phenomena may well 
complicate the analyses of the gonadectomy response. 

No attempt can be made to go into detailed genetic analysis of any of 
these phenomena at this time, but a few general statements can be made 
about these various animals. Not all hybrids produce exciting and un- 
expected pathologic conditions or physiologic changes and when these 
hybrids are backcrossed to the parental strains this situation continues 
to be true. Such is the case with the A and C3H mice. Therefore it 
may be said that the potential of these two strains in such combination 
does not yield indications for the development of any new strains. How- 
ever, strains CE, DBA, DE and C produce a wider variety of tumor types, 
(new phenomena not known in the parental strains and enhancement of 
tumor incidence' in one instance) so that these particular types of mice 
will be valuable in attempting to develop new strains known for some of 
these characteristics. 

Such then is one method whereby leads may be found in the continued 
search for valuable new types of mice that may be of help not only in 
meeting the needs of research but also as an aid to the clinician in the 
problems with which he is concerned. 


(1) Green, E. L.: The genetics of a difference in skeletal type between two inbred 

strains of mice (BalbC and C57blk). Genetics 36: 391-409, 1951. 

(2) Little, C. C.: Hybridization and tumor formation in mice. Proc. Nat. Acad. 

Sc. 25: 452-455, 1939. 
(8) Murray, W. S., and Little, C. C.: The genetics of mammary tumor incidence 
in mice. Genetics 20: 466-496, 1935. 

(4) Gorer, P. A., Lyman, S., and Snell, G. D.: Studies on the genetic and antigenic 

basis of tumor transplantation. Linkage between a histocompatibility gene 
and "fused" in mice. Proc. Roy. Soc. s.B, London, 135: 499-505, 1948. 

(5) Gordon, M.: Effects of five primary genes on the site of melanomas in fishes and 

the influence of two color genes on their pigmentation. Spec. Publ. New York 
Acad. Sc. 4: 216-268, 1948. 

(6) Fekete, E., Woolley, G. W., and Little, C. C: Histological changes following 

ovariectomy in mice. I. dba high tumor strain. J. Exper. Med. 74: 1-8, 1941. 

(7) Woolley, G. W., Fekete, E., and Little, C. C.: Effect of castration in the 

dilute brown strain of mice. Endocrinol. 28: 341-343, 1941. 

(8) Woolley, G. W., and Little, C. C: The incidence of adrenal cortical carcinoma 

in gonadectomized female mice of the extreme dilution strain. I. Observations 
on the adrenal cortex. Cancer Res. 5: 193-202, 1945. 

(9) : The incidence of adrenal cortical carcinoma in gonadectomized female 

mice of the extreme dilution strain. II. Observations on the accessory sex 
organs. Cancer Res. 5: 203-210, 1945. 

(10) : The incidence of adrenal cortical carcinoma in male mice of the extreme 

. dilution strain over one year of age. Cancer Res. 5: 506-509, 1945. 

(11) Woolley, G. W., Dickie, M. M., and Little, C. C: Adrenal tumors and other 

pathological changes in reciprocal crosses in mice. I. Strain DBA X Strain CE 
and the reciprocal. Cancer Res. 12: 142-152, 1952. 

(12) Unpublished data. 

(18) Woolley, G. W., Dickie, M. M., and Little, C. C: Adrenal tumors and other 
pathological changes in reciprocal crosses in mice. II. An introduction to results 
of four reciprocal crosses. Cancer Res. 13: 231-245, 1953. 

Journal of the National Cancer Institute 


(14) Dickie, M. M., and Woolley, G. W.: Spontaneous basophilic tumors of the 

pituitary glands in gonadectomized mice. Cancer Res. 9: 372-384, 1949. 

(15) Christy,. N. P., Dickie, M. M., Atkinson, W. B., and Woolley, G. W.: The 

pathogenesis of uterine lesions in virgin mice and in gonadectomized mice 
bearing adrenal cortical and pituitary tumors. Cancer Res. 11: 413-422, 1951. 

(16) Atkinson, W. B., and Dickie, M. M.: Further studies on the pathogenesis of 

uterine lesions in DBA X CE and reciprocal hybrid mice. Cancer Res. 13: 
165-167, 1953. 

(17) Atkinson, W. B., Dickie, M. M., and Fekete, E.: Effects of breeding on the 

development of ovarian, adrenal, and uterine lesions in DBA X CE and 
reciprocal hybrid mice. Endocrinol. 55: 316-325, 1954. 

Vol. 15, No. 3, December 1954 

Significance of Recent Developments in 
Nuclear Cytology and Cytogenetics of 
the Mouse u 2 

Elizabeth Fekete and Allen B. Griffen, 3 
Roscoe B. Jackson Memorial Laboratory, Bar 
Harbor, Maine 

I. Histopathological Studies 

Studies on a transplantable ovarian teratoma of a mouse which main- 
tained its pluripotency and is still capable of giving rise to differentiated 
tissues have been reported from this laboratory (1). 

In this paper we are presenting studies on another transplantable 
ovarian teratoma. This neoplasm, originally composed of both differ- 
entiated and undifferentiated embryonal elements, lost its power of 
differentiation after four subcutaneous transplant generations and at 
present is composed only of embryonic immature tissue. The sub- 
cutaneously transplanted tumor was successfully transformed into an 
ascites tumor by the method described by George Klein {2). It has been 
carried both subcutaneously as nodules and intraperitoneally as ascites 
tumor through several transplant generations. 

The neoplasm occurred in a 193-day-old C3H/Ks female in Dr. Kaliss' 
colony. The mouse had given birth to two litters. She was killed July 
16, 1952, at which time the abdomen was greatly distended. The left 
ovary was very large, measuring 40 X 20 X 10 mm. and the right ovary 
was about twice normal size. Many small nodules were scattered through 
the abdominal cavity, attached to different parts of the viscera, causing 
numerous adhesions. Some of the nodules had infiltrated the abdominal 
wall and were adhered to the skin; others penetrated the diaphragm and 
were attached to and infiltrated the lungs. The large left ovary had 
hemorrhagic areas and the central part was necrotic. Some of the small 
nodules contained black pigment, some were dense and solid, while others 
contained cysts and alveolar areas. 

Part of each ovary and several of the nodules were fixed, sectioned, and 
examined microscopically. The left ovary had only a narrow margin of 
non-necrotic tissue composed of several different types of tissues as well as 

i The second section of this paper was presented at the Symposium on 25 Years of Progress in Mammalian 
Genetics and Cancer, Bar Harbor, Maine, June 30, 1954. 

2 These investigations were aided by grants from the National Cancer Institute of the National Institutes of 
Health, U. S. Department of Health, Education, and Welfare. 

» The authors wish to express their gratitude to Miss Hope Otis and to Mr. Merrill C. Bunker for their capable 
assistance in all parts of this work. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

802 proceedings: SYMPOSIUM ON 25 YEARS of 

undifferentiated embryonic tissue. The right ovary contained some 
normal follicles and in the medullary part elements of the neoplasm which 
were embryonal (fi?. 1). The other nodules were composed of mixtures of 
well differentiated tissues and embryonal elements (figs. 2 and 3). The 
differentiated tissues included: nervous tissues composed mostly of 
neuroepithelial and neuroblastic elements, cartilage, bone, smooth muscle 
fibers and different types of connective tissues. Small cavities lined by 
ciliated respiratory epithelium, or tall columnar epithelium and goblet 
cells, resembling intestinal epithelium, or stratified squamous epithelium 
or pigmented epithelium were also present. Large cells and giant cells 
resembling trophoblasts composed part of a small nodule. Most of the 
tissues were richly cellular and mitotic figures were numerous. 

The neoplasm was designated E8156 and was diagnosed as a teratoma. 
As the left ovary contained the largest tumor this was considered to be the 
site of origin and the other nodules to be the result of dissemination and 

Several nodules of the neoplasm were cut into small pieces, mixed, and 
transplanted subcutaneously into five C3H/Fe mice at weaning age. 
Three of these were negative and two had large tumors at the site of trans- 
plantation within 3 months. These were retransplanted subcutaneously 
and the tumor has been carried this way through about 25 transplant 
generations. The rate of growth increased so that at present trans- 
plantation has to be performed about once in 3 weeks. The tumor grows 
almost 100 percent in males and females of strain C3H/Fe and their Fi 

Grossly the transplanted tumors were dense and hard for about three 
generations, after which they became very soft and hemorrhagic. Parallel 
with the gross changes there were alterations in microscopic appearance. 
There was a decrease in the amount of differentiated tissues; after the 
fourth transplant generation only undifferentiated embryonal elements 
were present, and the blood supply was very rich (figs. 4 and 5). Areas of 
giant cells resembling trophoblasts were often present (fig. 6). 

In an attempt to grow the neoplastic cells in the ascitic fluid, part of a 
subcutaneous growth of the fifth transplant generation was homogenized, 
diluted with saline and injected into the peritoneal cavity of C3H/Fe mice 
at weaning age. Parts of the same tumor were also transplanted sub- 
cutaneously. The intraperitoneal injection produced both numerous 
solid neoplastic nodules and a great increase of ascitic fluid in 4 weeks. 
About 2cc. of bloody ascitic fluid was drawn out with a syringe and re- 
transplanted intraperitoneally into new hosts. In these animals large 
numbers of neoplastic cells were present in the ascitic fluid but no solid 
nodules were found. The tumor has been carried by reinjecting ascitic 
fluid intraperitoneally through about 10 generations. The rate of growth 
increased and it is necessary now to reinject the ascitic fluid into new 
hosts about every 15 to 20 days. Ascitic fluid of C3H/Fe mice has been 
injected into the peritoneal cavity of 6 AKK/Fe mice and has produced an 

Journal of the National Cancer Institute 


increase in ascitic fluid containing tumor cells in all of them. It is planned 
to test the growth in several other strains. 

Ascitic fluid of C3H/Fe mice injected subcutaneously into the same 
strain produces local growth which is similar morphologically to the sub- 
cutaneously transplanted tumors. Any of the subcutaneously growing 
transplanted neoplasms can be readily converted into ascites form by 
homogenizing them and injecting them intraperitoneally. Cytologic 
studies of the tumor cells present in the ascitic fluid have been conducted 
on smear preparations by Dr. Griffen. 

II. Cytological Studies 

The successful completion of preliminary chromosome maps of the 
mouse by Slizynski (8, 4) and by Griffen (5) has shown that mouse 
cytogenetics, though difficult, has great promise of application in classical 
genetics and in tumor research. The present paper indicates that familiar- 
ity with the germinal chromosomes may enable the investigator to handle 
and to understand the nuclear phenomena of neoplastic cells with greater 
ease than has been possible in the past. The teratoma which has been 
described has provided some very good material for such studies. 

Both sections and teased smear-preparations of the nodular transplants, 
as carried by Dr. Fekete, revealed an abundance of cells in whose frequent 
mitotic figures the chromosomes were large and sharply defined. No 
satisfactory chromosome counts were obtained from these preparations, 
but the frequent appearance of polyploid cells and of multipolar spindles 
was observed. Upon the successful growth of the neoplastic cells in the 
ascitic fluid, samples of the material were prepared for high-magnification 
microscopy by the smear or squash method. The procedures may be 
summarized as follows : a) a drop of fluid, consisting of a rich cell suspen- 
sion, was placed upon each of 10 to 20 slides; 6) a drop of Sudan Black B 
stain-fixer, prepared according to a formula given by Cohen (6), was 
added to each slide and thoroughly mixed with the cell suspensions by 
stirring with a needle; c) cover glasses were placed on the preparations 
and each cover was firmly pressed and blotted with filter paper; d) all 
slides were made permanent by the vapor dehydration method of Bridges 
(7), followed by mounting in Euparal. 

Sudan Black smear preparations of the first ascitic generation revealed 
great numbers of cells as single bodies, 2-, 4-, and 8-celled groups, and 
other clusters (presumably clones) containing higher numbers of cells. 
Careful chromosome counts made from camera lucida drawings yielded a 
constant number of 38, or 2N minus 2 in terms of the normal mouse 
chromosome complement, for diploid cells. Polyploid cells, easily classi- 
fied as tetraploid and octoploid by means of direct chromosome counts, 
were of frequent occurrence. In all cases the ploidy was based on a 
haploid complement of 19, or N minus 1, so that all countable tetraploid 
cells, for example, contained no more than 76 chromosomes whereas 80 
should normally be expected. 

Vol. 15, No. 3, December 1954 
316263—54 37 

804 proceedings: symposium on 25 years of 

As further ascitic transplants were made, permanent smears were 
prepared; however, not all generations were included in the collection of 
material. Since the study of these slides is still in progress, comprehensive 
results cannot be presented at this time ; but the seventh ascitic generation 
has been given thorough analysis, revealing several striking characteristics. 
Countable diploid cells regularly showed more than 38 chromosomes. In 
all cases in which the metaphase plates were complete beyond doubt and 
showed no possibility of occlusions by superimpositions, the chromosome 
number was 42, indicating a gain of four chromosomes by some means. 
All polyploid nuclei were apparently based upon the new haploid number, 
N = 21, for all countable tetraploid and octoploid cells regularly showed 
more than 80 and more than 160 chromosomes, respectively. In addition 
to diploid, tetraploid and octoploid cells, the unexpected classes haploid 
(N = 21), triploid (3N = 63) and hexaploid (6N = 126) were found. 
The frequency of all cell types, as determined from a total metaphase 
count in one sample smear, was as follows: 

Degree of Chromosome Cell 

ploidy number count Percent 


























Tetraploid cells were found to have unusual reproductive features, which 
have not been observed in other cells, except for a single hexaploid cell to be 
mentioned below. Through normal bipolar mitosis a 4N cell may give 
rise to two 4N daughters with no unusual features observable in the spindle 
mechanism at anaphase. But in addition, 4N cells may form tripolar 
spindles and divide reductionally to produce one diploid (2N) and two 
haploid (N) daughters; or by means of tetrapolar spindles, 4N cells 
may divide reductionally into four haploid (N) daughters. While the 
old term ' 'reduction division" is hardly suitable for anything other than 
the meiotic phenomena seen in primary germ cells, there is nevertheless 
in these tumor-cell activities a very real reduction of chromosome number 
which is brought about through a separatory type of division without 
chromosome reduplication. It has not been possible to determine to 
what extent the reduced daughter nuclei may contain genetically com- 
plementary members of the chromosome complement, since the meta- 
phase chromosome morphology of the various daughters should need to 
be studied at great length; however, since haploid cells are frequently 
seen dividing, the haploid chromosome sets may be to some extent com- 
plete. In addition to the tetraploids, a single hexaploid (6N) cell has 
been seen dividing by means of a tripolar spindle to produce three polar 
chromosome clusters which were 3N, 2N, and N in approximate count. 

In meiosis the physical basis for chromosome segregation is synapsis. 
Since the phenomenon of synapsis has been observed frequently in tetra- 

Journal of the National Cancer Institute 


p]oid cells of the teratoma the haploid and other cells resulting from 
reduction may indeed approach genetic perfection in chromosome distri- 
bution, similar to that resulting from meiotic activities in germ cells. 
While the synapsis is apparently weak, it is at least strong enough to 
provide opportunity for segmental exchanges between chromosomes, for 
typical ' 'groups of four" are often found, sometimes several in a single cell. 
It has not yet been possible to determine whether the exchange figures 
result from crossing over or from translocation. Either of these phenom- 
ena would provide additional means for genetic variation in the tumor. 
Even though the tumor originated as tissue from the inbred C3H strain, 
there is no conceivable hindrance to gene mutation; crossing over can 
serve effectively in making new combinations of any mutants which may 
arise. Translocations, on the other hand, would provide directly for 
gene changes through position effects, and also would provide opportunity 
for segmental aneuploidy {i.e., duplication-deficiency phenomena involv- 
ing parts of chromosomes). 

From these observations it is apparent that the tumor, as presently 
grown in ascitic form, is capable of undergoing variation in the manner 
seen in sexually reproducing organisms. Mutation, synapsis, segmental 
exchanges, and segregation through reduction are in operation, without 
the necessity of sexual reproduction for survival. In the saprophytic or 
parasitic conditions of the ascites growth, variant cells are doubtless 
subject to selection as is indicated by the changes in chromosome number. 
The growth of the C3H tumor in AKR hosts further suggests evolution 
within the material. 

Hauschka and Levan (8) have presented extensive evidence that, in a 
number of tumor types, modifications of host-tumor specificities are 
based upon the predominance of 4N and other polyploid cells in the tumors. 
The activities of tetraploid cells in the teratoma suggest several possible 
mechanisms for such modifications. 


A spontaneously occurring ovarian teratoma which produced many 
secondary nodules in the peritoneal and pleural cavities is described. 

Transplanted subcutaneously this neoplasm showed a decrease in the 
amount of differentiated tissues; after the fourth transplant generation 
only undifferentiated embryonal elements were present. 

A homogenized subcutaneous tumor injected intraperitoneally produced 
an increase in ascitic fluid, containing neoplastic cells. 

The tumor is being maintained by subcutaneous transplantation and 
by reinjection of ascitic fluid intraperitoneally. 

The first ascitic generation, studied in Sudan Black smear preparations, 
had a diploid chromosome number of 38 or 2N-2. In the seventh ascitic 
generation the number had changed to 42. 

Polyploidy is common in the ascitic cells, tetraploid bodies appearing 
to be more numerous than others. 

Vol. 15, No. 3, December 1954 


Tetraploid cells, and presumably other polyploids, frequently show 
synapsis of chromosomes and figures which indicate segmental exchanges 
resulting from either crossing over or translocation. 

Through the formation of multipolar spindles, tetraploid cells may 
divide reductionally to produce diploid and haploid cells; the latter are 
cable of survival and are often found dividing normally. 

The teratoma exhibits the mechanisms for genetic variation and evolu- 
tion such as are found in sexually reproducing free-living organisms. 


(1) Fekete, E., and Ferrigno, M. A.: Studies on a transplantable teratoma of 

the mouse. Cancer Res. 12: 438-440, 1952. 
(#) Klein, G.: Comparative studies of mouse tumors with respect to their capacity 

for growth as "ascitic tumors" and their average nucleic acid content per cell. 

Exper. Cell Res. 2: 518-573, 1951. 
(5) Slizynski, B. M.: A preliminary pachytene chromosome map of the house 

mouse. J. Genetics 49: 242-245, 1949. 

(4) : Pachytene analysis of Snell's T(5; 8) a translocation in the mouse. J. 

Genetics 50: 507-511, 1952. 

(5) Griffen, A. B.: A late pachytene chromosome map of the male mouse. J. 

Morphol., 1954. In press. 

(6) Cohen, I.: Sudan Black B — a new stain for chromosome smear preparations. 

Stain Technol. 24: 177-184, 1949. 

(7) Bridges, C. B.: The vapor method of changing reagents, and of dehydration. 

Stain Technol. 12: 51-52, 1937. 

(8) Hauschka, T. S., and Levan, A.: Inverse relationship between chromosome 

ploidy and host-specificity of sixteen transplantable tumors. Exper. Cell 
Res. 4: 457-467, 1953. 

Plate 48 

Figure 1. — The right ovary of the mouse in which the teratoma occurred. The 
cortex contains normal ovarian follicles, the neoplastic medullary part is composed 
of embryonal tissue. X 200 

Figure 2. — A neoplastic nodule in the peritoneum of the same mouse showing heavily 
pigmented epithelium, neuroblasts, connective tissue and cartilage. X 150 

Figure 3. — Another neoplastic peritoneal nodule of the same mouse showing a small 
cyst lined by ciliated epithelium and another lined by stratified squamous epithe- 
lium. X 200 

Figure 4. — Subcutaneous growth of the sixth transplant generation showing well 
vascularized embryonal tissue. X 200 

Figure 5. — Subcutaneous growth of the fourth transplant generation showing 
embryonal tissue. X 425 

Figure 6. — Another subcutaneous growth of the fourth transplant generation show- 
ing large cells resembling trophoblasts. X 425 


Fekete and Griff en 
316263—54 — —38 


808 proceedings: symposium 

Plate 49 

Figure 7. — Smear of ascitic fluid showing neoplastic cells. Stained with Wright 
stain. X 500 

Figure 8. — Camera-lucida drawings of diploid metaphases from first ascitic genera- 
tions, showing the characteristic 38 chromsomes. X 650 

Figure 9. — Diploid metaphase from first ascitic generation; Sudan Black smear 
photographed with phase contrast. X 1,900 

Figure 10. — Octoploid metaphase from seventh ascitic generation, showing more 
than 160 chromosomes; Sudan Black, phase contrast. X 1,600 

Figure 11.— Tetraploid anaphase from seventh ascitic generation, dividing by means 
of a tripolar spindle into two haploid chromosome clusters (above) and one diploid 
(below); fresh Sudan Black smear, phase contrast. X 1,200 

Figure 12. — Tetraploid metaphase from seventh ascitic generation showing three 
exchange configurations (arrows); Sudan Black smear, phase contrast. X 1,900 






* J» 



m f| 



i: .« 

, : « : 'ill 






2l %*/ _ 

^"^*» 'P'* *^T— -tit" 

Fekete and Griffen 



Dr. Donald F. Jones, The Connecticut Agricultural Experiment Station and 
Yale University, New Haven, Conn. 

Long before the rediscovery of Mendelism, chromosome aberration had been asso- 
ciated with abnormal growth. It is now generally considered that these irregularities 
in chromosome behavior are the result and not the cause of the changes in growth. 
Visible chromosomal aberrations have obscured minute changes that initiate the first 
departure from normal. So far, these small rearrangements have been detected only 
by the use of genetic markers that produce alterations visible in single cells. Plant 
tissues are generally more favorable material in which to see these changes, as plants 
have cell walls and many genes produce nondiff usible cell products. Breaks and realign- 
ments of chromatin have been shown to initiate growth changes (1) . 

Recent studies by McClintock, Brink and others have shown position effects result- 
ing from migrating pieces of chromatin in maize and other plant material. These 
result in unstable gene effects clearly visible in variegated colors. So far these mutable 
genes have not been associated with growth changes. Since there is definite evidence 
that genes control growth in many ways, unstable effects would be expected when 
migrating chromatin pieces come in contact with growth-regulating regions. 

The excellent cytologic and cytogenetic studies that Dr. Griffen has reported give 
additional evidence for the importance of chromosome rearrangements in abnormal 
growth. The detailed chromosome maps will be an important aid to future studies. 

Evidence from all sources combines to show that organisms are a product of the 
genes, the cytoplasm and the environment, both external and internal. A normal 
genome-plasmone-environ interaction is dependent upon a well balanced system re- 
sulting from countless generations of selecting under all possible environmental condi- 
tions. There is good evidence that a failure of this genome-plasmone relation in the 
normal environ is the beginning of species separation (2) . 

Abnormal growth that is irreversible has much in common with species formation, 
since it is a new kind of tissue that has been removed from normal control. There is 
every reason to expect that the same factors involved in species separation will also be 
found to operate in the initiation of tumor formation. 


(1) Jones, D. F.: Nuclear changes affecting growth. Am. J. Botany 27: 149-155, 


(2) : The cytoplasmic separation of species. Proc. Nat. Acad. Sc. 37: 

408-410, 1951. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Session VIII. Genetic Techniques in 
the Study of Cancer: New Approaches 

Chairman, Dr. Clarence C. Little, Director, 
Roscoe B. Jackson Memorial Laboratory, Bar 
Harbor, Maine 

Speaker: Dr. Lloyd W. Law 

Studies on Transformations in Leukemic Cells of the Mouse 
Discusser: Dr. Arthur G. Steinberg 

Speaker: Dr. Sewall Wright 

Summary of Patterns of Mammalian Gene Action 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

Introduction: Session VIII 
Dr. C. C. Little, Chairman 

It might be of interest to mention at this time, one or two principles or guiding 
"lines of thought" that I have found helpful as reference points to which one might 
from time to time possibly return during studies on growth. 

The first of these principles is the great truth that animal cells in the organized body 
possess a far greater latent power of growth than the student of a particular problem of 
normal or abnormal growth is likely to recognize. The most eloquent and enduring 
effect of the latent power of growth is the unbroken history of the protozoa, which has 
shown us that the division rate of the healthy, free animal cell has not been slowed 
down through the geological periods. The germ cells of mammals, as another example, 
are liable to be underestimated. Also the organs that give rise to these germ 
cells should, I think, be viewed with more respect as definite centers of latent-growth 
power. The process of regeneration is an extraordinary evidence that many cells 
otherwise controlled can, when challenged by trauma or by other experience, rise to 
the occasion without the addition of any chemical substance, without any experimental 
change. That they are able to replace structures is striking evidence of the latent 
power of growth. The ordinary repair processes of the body which we take for granted 
are really evidence of immense latent power of growth in the differentiated cells. 
Because we see these cells so definite, so diversified, we are apt to think of them as being 
permanently limited. This is not the case. In other words, a great latent power of 
growth is submerged in the higher organism, but is not lost. The power of replacement 
of tissue is still less sensational, but is tremendously impressive. It is the expression 
of enormous latent power of growth. A very unhappy but striking demonstration of 
latent power of growth is the origin of neoplasia from cells in differentiated organisms. 
Latent power of growth is therefore one great principle which we have largely neglected 
in our own thinking. 

It is well to realize that what we call the "normal" cell of the mammal is really quite 
"abnormal," from the point of view of successful primitive animal cells. The protozoa 
are much more typical; cancer cells are much more efficient. They outlive, and out- 
last the differentiated environment around them. The differentiated cell of the 
mammal is "usual," but it is not a normal self-sufficient biological cell in comparison 
with those that are expressing the birthright of every healthy animal cell, which is 
unlimited growth. 

The second great principle is that of balance and unbalance. This principle has also 
been largely overlooked because we are all much more comfortable intellectually if we 
deal with end products of a reaction, rather than the reaction itself. If we are geneti- 
cists we like the finished character or the chromosome; if we are pathologists we like 
to have microscopic cells to study and classify. We all like definite things, because 
our intellects are definite; and yet the whole process of life, the characteristic thing 
that makes it baffling and bewildering, is that it is a dynamic function. It is something 
which is moving and it tends to move on a basis of balance. It can move around and 
from the "center" of balance to meet various "challenges." The body does this 
normally in every body function which has any cyclic phase whatever. Any sequence 
of buildup and destructive phases is a cycle. It is a question of balance, and the cell 
must come back, or the tissue, or the organ, must come back to a "resting" or balanced 
stage which is health, which is normal, which is in a very real sense a balance. Now the 
question of unbalance becomes critical. When the pendulum is swung so far by a 
challenge that it sticks at a new center of function, a new center of balance is established. 
Sometimes this is to the advantage, and sometimes to the disadvantage, of the com- 
plicated organism in which the shift occurs. We know that we can upset balance in 
a great many ways. By irradiation, by temperature changes, for example, we can 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

816 proceedings: symposium 

challenge the balance of the organism. We also know it is challenged by various 
carcinogens and also as you heard today by the process of hybridization. 

When you bring together in a cross, two germ cells that are alike, or as nearly alike 
as you can get them by the process of inbreeding, it is obvious that the resulting 
systems of physiology and morphological structures worked out during development 
tend to coincide. The two parental tendencies maintain a balance, repeat each 
other's rates and tempos, so that the organism is not greatly challenged. But when 
you bring together two systems that differ in tempo, either the total growth tempo or 
periodic or local tempo, they challenge each other. There follows an increase in 
tumors, an increase in mutation after hydridization. This may have an important 
bearing on the process of heterosis. If one admits a latent power of growth bringing 
together two unlike cell-potential systems, one system may have the gene to "stop" a 
certain growth process at a certain time and the other may lack it. One then has 
produced the maximum challenge to growth control, the maximum weakening of the 
"control" of this latent power of growth. Heterosis may be due to more unbalances 
and less ability to stop growth processes at the same time and in the same manner. 
One gets the summation of the growth potentials of both parents, and because they 
don't coincide and are different, they compete with each other. Heterosis might then 
depend upon the unbalance between the tempos of the control of growth in the con- 
flicting systems. 

The principle of unbalance applies all the way down the size scale between organs, 
between tissues, between cells in the tissue, between the nucleus and cytoplasm of the 
cell, between chromosomes within the nucleus, and between genes, within the chromo- 
some, and finally in the conformation of the genes themselves. 

We have wondered why the same agents produce neoplasia and mutation. It 
would seem probable that it is the unbalance produced by these agents which, in 
cellular terms, produces a break in control within the cells as a unit. This results 
in more rapid cell division. In the gene, the agent breaks the conformation or organi- 
zation of the gene, producing unbalance and later the establishment of a new balance 
in the chemical structure of that gene, a new relationship between the atoms. It is 
probable that mutation is not neoplasia, or neoplasia mutation, but that both processes 
depend upon establishment of a new balance, one intracellular, and the other intra- 
genic. This consideration of latent growth potential and balance versus unbalance 
may help to bring certain scientific disciplines together, and may make us able to 
exchange a great deal of evidence on dynamic processes which though different in 
their end products, may have a common origin, a common basic philosophy back of 

This afternoon we are indeed very privileged to hear two papers. The first of these 
is going to deal with transformations in leukemic cells by Dr. Lloyd Law of the National 
Cancer Institute. The second one is going to be a summary and digest of the subject 
matter of this symposium by Dr. Wright. First, then, I should like to call on Dr. Law 
to speak on the transformations of leukemic cells. 

Studies on Transformations in Leu- 
kemic Cells of the Mouse * 

L. W. Law, National Cancer Institute, 2 Bethesda, 

Two groups of compounds, classed as antimetabolites, have been used 
to a great extent in the treatment of acute lymphocytic leukemias of 
children and in laboratory investigations employing this morphologic 
form of leukemia in mice: 1) folic-acid antagonists, particularly those 
with a 4-amino substituent, and 2) purine antagonists. Folic-acid antag- 
onists have been found, in our laboratory, to be antileukemic agents to 
each and every lymphocytic leukemia tested. On the other hand, purine 
antagonists are effective antileukemic agents in some, but not other, 

These metabolic antagonists are of interest for several reasons: 1) They 
are the most effective of known antileukemic agents, 2) selectivity of 
action is, in particular cases, striking, and, 3) the inhibitory effect can 
be shown to be of a competitive nature, in certain instances, affording a 
better means of elucidating the metabolic reactions involved in therapeutic 

The failure, after a period of time, to achieve remissions in patients 
with A-methopterin and 6-mercaptopurine is a common observation. In 
the experimental leukemias it has been shown that these failures result 
from the development of transformations in the population of leukemic 
cells to resistance and/or dependence of various degrees. 

It is the purpose of this report to consider the experimental production 
of these transformations, some characteristics of the transformed leukemic 
cells, and to investigate the manner of origin of these variant cells. Some 
basic information obtained involving the mechanisms of these phenomena 
will be given. 

Folic-Acid Analogs 

Both types of transformation, to resistance, wherein leukemic cells 
grow optimally either in the presence or absence of antagonist or, to 
dependence, wherein the cells grow optimally only in the presence of 
antagonist, have been obtained using the 4-amino-substituted folic 
analogs, 4-amino PGA (aminopterin) , 4-amino-N 10 -methyl PG (A-methop- 
terin), 4-amino-9-methyl PGA (A-ninopterin), and 4-amino-9,N 10 -methyl 
PGA (A-denopterin) 3 (1-3). See text-figure 1 and table 1. 

i Presented at the Symposium on 25 Years of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 30, 1954. 

2 National Institutes of Health, Public Health Service, U. S. Department of Health, Education, and Welfare. 

3 All folic-acid analogs have been supplied through the courtesy of the Lederle Laboratories Division, American 
Cyanamid Co. Purine analogs have been supplied by Dr. George H. Hitchings, Wellcome Research Labora- 


Journal of the National Cancer Institute, Vol. 15, No. S, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 



— Ts~l Til 

(40) (40) 





Text-figure 1. — Showing origin of various transformed sublines discussed in text: 
AN-R (resistant to A-ninopterin) , AD-R (dependent on A-denopterin) , AM-D 
(dependent on A-methopterin), 8-AG-D (dependent on 8-azaguanine) , AM-R 
(resistant to A-methopterin), 8-AG-R (resistant to 8-azaguanine), 6-M-R (resistant 
to 6-mercaptopurine) . Hatched squares represent resistance; darkened squares, 
dependence. The control line, represented by circles, has been carried through 240 
consecutive transfers and remains sensitive to the antifolic and antipurine com- 
pounds. Certain transfer lines have been discontinued as shown; others have been 
carried serially through the number of transfers designated below the square. 

Table 1. — Transformations in leukemic cells of several transplantable lines 

Line of 

Strain of 

Type of transformation 

Antagonist used 




















































Variant sublines of transplantable acute lymphocytic leukemias are 
obtained usually with ease following consecutive serial transfers in mice 

Journal of the National Cancer Institute 



receiving either 1) near-maximum tolerable levels of antagonist, or 2) 
periodic increases in the level of antagonist. Once transformation has 
been achieved the variant lines retain their characteristics; no reversion 
to sensitivity has been observed or changes from one type of transforma- 
tion to another. These characteristics are maintained in the absence of 
antagonist used to achieve the transformation. The character is thus 
shown to be stable, irreversible, and heritable. 

One particular subline of an acute lymphocytic leukemia, L1210, the 
AM-D subline, dependent on folic antagonists for optimal growth, has 
been of interest in determining some of the physiologic characteristics of 
these transformed cells. 

Characteristics of Leukemic Cells Dependent on Folic Analogs jor Optimal 


The AM-D variant line has now been carried through more than 200 
consecutive serial transfers in strain DBA/2 mice. Optimal growth is 
obtained in the presence of 4-amino-N 10 -methyl PGA (A-methopterin) , 
the analog used to develop dependence. The behavior of these trans- 
formed cells is shown in table 2 in comparison with the sensitive, control 
line, using the criterion of localized growth of lymphomatous tissue, and 
in table 3 the behavior in the development of generalized leukemia follow- 
ing transfer of a standard dose of leukemic cells in Locke's solution (8 X 10 5 
cells) . It is of interest to note that even at a dosage level of A-methopterin 
as high as 12 mg. per kg. (total dose, 48 mg./kg.) an inhibition of localized 
growth of lymphomatous tissue of 50 percent occurred but infiltration into 
lymph nodes and spleen was moderate. This indicates at least a 50-fold 
increase in tolerance to this antagonist, since it requires 0.25 mg. per kg. 
(total dose 1.0 mg./kg.) of antagonist to inhibit localized growth of the 
sensitive cells to a similar degree. 

A specific cross-dependence which is characteristic for all such trans- 
formations is shown in table 4. Any 4-amino-substituted folic antagonist 
is capable of providing for optimal growth. Some of the so-called "weak" 
antagonists, notably N 10 -methyl PGA and 9-methyl PGA, though lacking 

Table 2. — Dependence in leukemic cells* of the AM-D 
subline oj leukemia LI 2 10 

of mice 

dosage (mg./kg.) 

Mean weight 
tous tissue 










212. 7 






•Transfer generations 98-108 of sensitive cells and 33-39 of dependent cells used. 
Vol. 13, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 



+ + + 

+ 1 



+ + + 1 

+ 1 





+ + 

+ 1 


+ + + 

+ 1 


+ + + + 

+ 1 




+ + 

+ 1 


+ + H — h 

+ 1 


+ + + + 

+ 1 



tH lO 



CO 10 


00 (M 00 



W t-i<N 






OS 03 00 

<M <N 



<N OH t- 

00 >o 



CO i-i CO 








iH 02 

a +3 


CO lO 


O5t>00 00 

10 10 

c3 >, 


5 -1 


^ CO 












1^05 00 CO 




1-H T— 1 


Tj< 1-1 


i-H <N 

<N lO 


0<N <NCO 

1-H CO 

i— 1 1— 1 1— ( i-H 


b- b- 






S a> B 





H £ w 


- r 




<M OO00 






















• v > 
















a ^ 

> + 

r + 

9 5 + 

Journal of the National Cancer Institute 



Table 4. — Comparative sensitivity of A-methopterin-dependent (AM-D) and sensitive 
leukemic cells to several antileukemic agents 



Individual Total 

Inhibition index* 

Sensitive Dependent 

8-Azaguanine , 





Amino-an-f ol 

















. 17 
. 10 



♦Inhibition index- 

Mean wt. tumor-treated 

Mean wt. tumor-controls 

antileukemic activity, are also able to provide for approximately 50 percent 
optimal growth of these dependent cells. Such antileukemic compounds 
as 8-azaguanine, aZpAa-peltatin, TEM and 6-mercaptopurine, on the other 
hand, show independence of action, in being carcinostatic for either the 
dependent or sensitive sublines. Two compounds, cortisone and a 
purine antagonist, 2,6-diaminopurine, were ineffective in either the 
dependent or sensitive subline in this study. 

The 4-amino-substituted folic analogs appear to inhibit leukemic-cell 
growth by antagonism of folic acid (PGA) or citrovorum factor (CF) 
since either of these compounds will prevent antileukemic action of this 
class of agents (4) . Since it appears that CF, on a weight basis, is more 
effective than PGA in reversing the effects of folic analogs, it was of 
interest to study the effects of this metabolite on the growth characteris- 
tics of the sensitive and dependent lines of leukemia L1210 and to deter- 
mine the blocking effect of CF on 1) the antileukemic action of A-methop- 
terin on sensitive leukemic cells and 2) the optimal growth-promoting 
capacity of this antagonist on dependent leukemic cells. It may be seen 
by reference to table 5 that partial reversals of both the antileukemic 
effect in the sensitive line and of the growth-promoting effect in the 
dependent line were obtained. Folic acid (PGA) was also found to give 
similar reversals. The ratio of analog to metabolite (PGA) for maximum 
effect was approximately 1:15 and the ratio was the same for sensitive 
or dependent cells, provided that PGA was always given prior to ad- 
ministration of analog. PGA and CF were found not to influence the 
growth characteristics of either sensitive or dependent cells, under the 
conditions of these experiments. 

No changes in morphology, antigenicity or transplantability have been 
observed in the several variant forms of resistant or dependent cells 
developed in leukemia L1210 or the other transplantable leukemias em- 
ployed, although it is to be noted that specific differences in growth rates 
do exist. Leukemic cells of the sensitive and of the transformed sublines, 
as observed in localized growth or in infiltrations into spleen, liver, 
lymph nodes or in the peripheral blood are morphologically indistin- 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 

guishable. Attempts to detect antigenic differences by complement- 
fixation tests have been unsuccessful. 

Possible Mechanisms of Resistance and Dependence 

Resistance of leukemic cells to folic analogs may be the result of 1) a 
lowered PGA requirement accompanied by a much greater capacity to 
convert PGA to utilizable CF, 2) an increased ability to detoxify the PGA 
antagonist, 3) a more efficient utilization of CF due to changes in permea- 
bility of the cell, or, 4) the ability of transformed cells to convert antagonist 
to PGA or CF by one of several methods: deamination, demethylation, 
etc. In the case of Streptococcus jaecalis, resistant to folic antagonist, it 
appears that this strain has a lowered requirement for PGA and a much 

Table 5. 

-Citrovorum factor (CF) and the effects of A-methopterin on sensitive and 
dependent (AM-D) leukemic cells of leukemia LI 2 10* 


of mice 





weights f 

Dependent (AM-D) 


+ CFJ 




+ CFJ 



"[Data from Law (2).] 

tFor convenience, optimal growth in both groups considered as 1.0. 

JDosage of A-methopterin in mg./kg.: citrovorum factor dosage in units per kg. 



greater capacity to convert PGA to CF than the antagonist-sensitive 
strain so that significant inhibitions of growth are obtained only with very 
high concentrations of antagonist (5). Evidence is now at hand to indi- 
cate that resistance to A-methopterin by S. jaecalis also involves an altered 
permeability of the cells to antagonists which greatly reduces the acces- 
sibility of the susceptible enzyme system to antagonist. 

It has been reported recently (6, 7) that a folic-antagonist-resistant 
strain of S. jaecalis was able to use aminopterin and related analogs for 
growth, in contradistinction to the sensitive strain, by converting the 
analogs to PGA or CF. Similarly, Kidder et al. (8) observed that the 
protozoon, Tetrahymena, was able to use aminopterin (and methopterin) 
in growth processes, suggesting that this organism also possessed enzymes 
capable of deaminating and demethylating these antagonists. It has been 

Journal of the National Cancer Institute 


found by Nichol et al. (5), however, employing paper chromatographic 
techniques, that the folic-acid antagonists, which are presently available, 
contain sufficient PGA and pteroic acid as impurities to account for the 
apparent utilization of these analogs by S.faecalis and Tetrahymena and 
that the explanations given above are not valid. 

In considering explanations for the dependence characteristic, it is ap- 
parent that many of the suggestions pertaining to resistance are unlikely. 
For example, the ability of leukemic cells to detoxify the antagonist or 
to convert antagonist to PGA or CF would explain the phenomenon of 
resistance but not dependence. Preliminary trials by Nichol (9) on the 
ability of our L1210 A-methopterin-dependent (AM-D) leukemic cells to 
convert PGA to CF indicate that the dependent leukemic cells are less 
active than sensitive cells in this respect, in contrast to the results ob- 
tained with S. faecalis. 

Also, it is quite likely that differential absorption of folic analog or PGA 
is not related to the dependence phenomenon. Preliminary trials by 
Skipper et al. (10) using C u -labeled PGA and A-methopterin show C 1 * 
contents of sensitive and dependent L1210 leukemic cells to be of the 
same order. Recent evidence by Kieler and Kieler (11) on the action of 
A-methopterin on leukemic cells in vitro indicates the possibility of differ- 
ences in intracellular distribution of PGA and its antagonists. There 
are no definitive data available at present relating to such distribution. 

It is possible that the dependent leukemic cells described here have 
acquired the ability to use folic analogs without conversion to PGA or CF, 
employing a different mechanism for the synthesis of nucleic acids than 
that suggested as occurring normally. Certain preliminary data are 
available relating to this interpretation. As mentioned previously, it is 
evident that the 4-aminopteroylglutamic acids act by competing with 
PGA or CF. It appears that PGA is converted to CF which is associated 
with enzymes concerned with transfer of single carbon units. Folic an- 
tagonists compete with formed CF, the end result of this antagonism is an 
inhibition of nucleic acid synthesis as well as other biochemical processes. 
It has been found (12), using NaFormate-C 14 (a precursor of the 2- and 
8-carbon atoms of nucleic acid purines) that the folic-acid antagonist, 
A-methopterin, inhibits de novo synthesis of DNA and UNA purines of the 
sensitive cells and viscera of mice bearing these cells while the analog 
more than doubles the rate of de novo DNA and RNA synthesis in the 
dependent leukemic cells, profoundly inhibiting the nucleic acid synthesis 
in the viscera of mice bearing these transplanted cells. These results are 
shown in table 6. 

Similar results were obtained from in vitro studies of sensitive and 
dependent leukemic cells from the same sources. A-methopterin at con- 
centrations of 0.01 mg. per ml. strongly inhibited the incorporation of C 14 - 
formate into the protein and purine pentose nucleotides of sensitive cells. 
A-methopterin at much higher levels was found to be ineffective on the 
in vitro incorporation of P 32 04 in sensitive and transformed cells. These 
results indicate that the effect of folic analogs is not an over-all inhibition 

Vol. 15, No. 3, December 1954 
316263—54 39 


proceedings: SYMPOSIUM ON 25 YEARS of 

Table 6. — Incorporation of C u -formate into nucleic acid moieties of viscera and 
A-methopterin-sensitive and dependent (AM-D) tumor masses of leukemia L1210* 





Specific activities (nc/mole C) 










Sensitive. . 
Sensitive. . 

Sensitive. . 
Sensitive . . 

Dependent . 
Dependent . 

Dependent . 
Dependent . 

Viscera . 
Tumor . 

Viscera . 
Tumor . 

Viscera . 
Tumor . 

Viscera . 
Tumor . 


























A-methopterinf . 
A-methopterinj . 






A-methopterinJ . 
A-methopterinj . 




*Data from Skipper, Bennett and Law {12). 

fA-methopterin (3 mg./kg.) immediately before HC 14 OONa injection on 7th post- 
inoculation days. 

JA-methopterin (3 mg./kg.) on 3d, 5th, and 7th days. HC 14 OONa on 7th day. 

of tissue metabolism (18), but is probably directed against limited enzyme 

It should be pointed out that, although the available folic antagonists 
contain certain contaminants which are growth factors, it is not neces- 
sarily a complicating factor in the production and development of trans- 
formations to resistance and dependence. Contrariwise, it would appear 
that the presence of PGA in the antagonist used in S. faecalis experiments 
aided in the selection of resistant mutants. It should be recognized, how- 
ever, that some confusion in the interpretation of results has arisen. 

Though the mechanism of dependence, in biochemical terms, is far 
from a solution, some definite leads have been established. Additional 
evidence, which tends to support the thesis that dependent leukemic cells 
employ a different mechanism for the synthesis of nucleic acids than that 
suggested as occurring normally, is to be found in the use of weak folic 
antagonists. Although N 10 -methyl PGA is found not to inhibit the growth 
of sensitive leukemic cells it does provide for approximately 50 percent 
optimal growth of dependent cells (L1210 AM-D). It has been de- 
termined that lymphomatous tissue or spleen obtained from (L1210 
AM-D) mice apparently does not utilize this compound as a precursor for 
CF (9). 

Purine Analogs 

Since the report of Kidder et al. in 1949 (14) showing the cancerostatic 
effect of a triazolopyrimidine analog of guanine, 8-azaguanine, on certain 
adenocarcinomas and a leukemia in mice, this compound has been studied 
rather extensively. It has proved to be a useful and interesting tool in 
investigations of cellular biochemical reactions. Although specific and 
definite inhibitory action has been noted for a fairly wide range of mor- 

Journal of the National Cancer Institute 



phologic forms of neoplasm (15-17), it is equally clear that it is entirely 
inactive against other neoplasms in the mouse and rat (15, 16). This 
dichotomy of response to 6-mercaptopurine and thioguanine has been 
observed also in a study of a spectrum of acute lymphocytic leukemias of 
the mouse (23). An explanation for response or lack of response to 8- 
azaguanine has been given by Gellhorn and co-workers. 

In certain acute lymphocytic leukemias of the mouse (17), particularly 
the transplantable leukemia L1210, a definite, regular and reproducible 
inhibition of growth results following parenteral administration of the 
guanine antagonist at nontoxic levels far below the maximum tolerable 

Transformations to resistance and to dependence have been obtained 
in leukemic cells of Line L1210 by consecutive serial passage in DBA/2 
mice receiving near MTD levels of 8-azaguanine (see text-fig. 1). The 
dependent line (8-AG-D) was developed from the 100th transfer of sensitive 
cells and a resistant line from the 175th transfer (18). 

Transformations to Resistance and Dependence Using 8-Azaguanine 

Table 7 shows the characteristic of dependence in the transformed cells. 
Optimal growth, as determined by localized growth of lymphomatous 
tissue, was obtained at dosage levels of 8-azaguanine as high as 150 mg. per 
kg. (total dose, 1,200 mg. per kg.). At this level complete inhibition of 
growth is observed in the sensitive line. Leukemic death, following 
intraperitoneal transfer of cells, also strikingly reflects the dependence 
characteristic. The mean survival time of mice bearing the dependent 
subline was 15.8 ± 0.45 days. If these mice are given 8-azaguanine 
(75 mg. per kg. X 8) they die earlier, 12.1 ± 0.23 days with a florid 
leukemia (see table 9). In contrast DBA/2 mice bearing the control, 
sensitive subline die at 7.9 ± 0.06 days and when given 8-azaguanine 
parenterally, at 12.4 ± 0.20 days. 

This dependent subline has now been carried through 100 transfer 
generations, since emergence of the trait, and has retained its characteristic 
response without evidence of reversion. 

Table 7. — Effect of 8-azaguanine on sensitive and 8-azaguanine- 
dependent (8-AG-D) lines of leukemia LI 210* 

Transfer line 

of mice 


Tumor wt. at 
9 days 




f 24f 

[ 89 




591. 1 

538. 6 ± 42. 1 

240.4 ± 21.8 


f 10 


[ 54 




16.8 ± 2.2 

775.6 ± 30.2 

•[Data from Law (18).] 

fTransfer generations 107-144 of the sensitive line and 7-44 of the dependent line used. 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 

Cross-dependence on other purine analogs has been demonstrated in 
this 8-azaguanine-dependent line. 6-Mercaptopurine and thioguanine 
(both moderately carcinostatic) , 8-azaxan thine, and 2,6-diaminopurine 
(ineffective for sensitive leukemic cells) all provide for 50 percent or more 
optimal growth of the dependent line. The folic antagonist, A-methop- 
terin and TEM are inhibitory for the dependent cells as well as the 
sensitive. A striking sensitivity to folic analogs of the dependent cells, 
as well as of other transformations produced by purine analogs has been 
noted and will be discussed later. 

L1210 leukemic cells transformed to resistance, using 8-azaguanine, 
grow optimally in DBA/2 mice with or without this antagonist, succumbing 
from leukemia at 10.1 ±0.15 days. Cross-resistance to all other purine 
analogs has been demonstrated (see table 8) but these resistant cells 
remain sensitive to other unrelated compounds such as folic analogs and 
TEM. This line has now been carried through 85 consecutive serial 
passages in DBA/2 mice, retaining its characteristics. 

Resistance to an Adenine Antagonist, 6-Mercaptopurine 

The adenine analog, 6-mercaptopurine, has been shown to act as a 
purine antagonist in the metabolism of Lactobacillus casei (19). It has 
also been shown to be a unique inhibitor of Sarcoma 180 (20) and of 
certain mammary adenocarcinomas (10). Limited clinical trials of this 
compound in advanced leukemia of children have been encouraging (21). 
As with 8-azaguanine this compound has been shown to give definite, 
regular and reproducible inhibition of leukemic-cell growth in certain 
lymphocytic leukemias but not others (22, 23). Striking increases in 
survival time occur in leukemia L1210. The mean survival time of mice 
bearing the sensitive subline of leukemic L1210 was 7.9 ±0.06 days and 
an increase in survival of 87.3 percent, to 14.8 ±0.18 days was obtained 
using this antagonist within the total dosage range of 250 to 1,200 mg. per 
kg. The effects obtained at the higher dosage levels (near MTD) were 
within the same range as those obtained at lower levels, 300 to 600 
mg. per kg. 

A resistant line was procured starting with the 200th transfer of the 
sensitive line. The resistance characteristic was apparent after 5 consecu- 

Table 8. — Influence of several purine antagonists on variant sublines 
of leukemia LI 210* 










+ (100%) 

+ ( 80%) 

+ ( 80%) 

-( 60%) 
-< 50%) 
-( 50%) 
-( 40%) 


* +=inhibition; — -support of growth (cross-dependence); 0=no influence (cross-resistance). 
t 8-Azaguanine-dependent. 
J 8-Azaguanine-resistant. 
§ 6-Mercaptopurine-resistant. 

Journal of the National Cancer Institute 


tive transfers in DBA/2 mice receiving 75 mg. per kg. X 7 dosage levels. 
No influence of 6-mercaptopurine could then be demonstrated in this line, 
test mice with and without antagonist dying at 9.9 ±0.10 days. Cross- 
resistance, similar to that observed in the 8-azaguanine-resistant line, was 
found using other purine analogs (table 8) but sensitivity to TEM and 
A-methopterin was evident. A considerably increased sensitivity to the 
folic antagonist was characteristic. 

Mode of Action of Purine Antagonists 

There appears to be little doubt that the purine antagonist 8-azaguanine 
is incorporated into nucleic acids. This has been demonstrated by 
Heinrich et al. (24) in the protozoon, Tetrahymena, by Mitchell et al. (25) 
for viscera of mice and more recently, using finer techniques, for mouse 
viscera and tumor tissue by Skipper (10). The incorporation is for the 
most part in RNA and in relatively small amounts. There appears to be 
little doubt also in Tetrahymena, which requires preformed guanine, that 
the guanine antagonist acts strictly as an antimetabolite; physiological 
guanine, or its nucleotide, reversing in competitive fashion the growth- 
inhibiting capacity of 8-azaguanine. Evidence for a clear-cut metabolite- 
antimetabolite relationship in mice or other mammals is not yet at hand. 
Guanine or guanylic acid has been shown to reverse the carcinostatic 
effect of 8-azaguanine, as determined by leukemic deaths in mice (17, 26). 
In our own observations with the 8-azaguanine-dependent leukemia, 
guanylic acid regularly interferes with the growth-promoting capacity of 
8-azaguanine more effectively than another ribotide, adenylic acid. It 
has not been determined whether this is done competitively. On the 
other hand, Gellhorn et al. (27) have observed, in rabbits bearing the 
Brown-Pearce carcinoma, that the carcinostatic effects of 8-azaguanine 
are more easily reversed by the nucleosides and nucleotides of adenine. 
This suggests that 8-azaguanine is converted first to adenine prior to 
conversion to a riboside. 

Extensive attempts in our laboratory (23) to reverse the carcinostatic 
activity of 6-mercaptopurine by physiological purine bases have been 
relatively unsuccessful, although on occasion reversals have been obtained 
particularly with hypoxanthine. In Lactobacillus casei any of the four 
physiological purines will prevent the inhibition produced by this com- 
pound, easily and competitively (19). The negative outcome of reversal 
experiments, using 6-mercaptopurine, does not necessarily mean that its 
mode of action is different in these experimental animals as compared with 
Lactobacillus but that the techniques employed in the complicated system 
in experimental animals are not adequate or that some metabolite other 
than the four physiologic purines must be supplied. 

Some suggestive preliminary data obtained through a study of nucleic 
acid metabolism of sensitive and dependent (8-azaguanine) leukemic cells 
of leukemia L1210 are at hand. 8-Azaguanine 2-C 14 has been found to be 
incorporated into the RNA of sensitive leukemic cells at levels 100 times 
the incorporation of this purine antagonist into dependent cells (27). 

Vol. 15, No. 3, December 19S4 

828 proceedings: SYMPOSIUM ON 25 YEARS of 

This may be considered as evidence that fixation of 8-azaguanine in nucleic 
acids may be related to the carcinostatic activity of the compound. Low 
levels of incorporation of the purine antagonist 2,6-diaminopurine (as 2,6 
DAP-2-C 14 ) in dependent cells has also been found (10) paralleling the 
results with 8-azaguanine, whereas the utilization of thymine and guanine 
(as 2-C 14 products) are of the same order of magnitude in sensitive and 
dependent cells, a relatively low incorporation of guanine being observed. 
Since 2,6-diaminopurine is known to be readily converted to nucleic acid 
guanine (28) these results indicate a difference in capacity of the two 
types of cells to utilize this compound as a source of guanine. 

The observations discussed here concerning metabolism of sensitive and 
transformed leukemic cells suggest that differences in nucleic acid metabo- 
lism may exist. Attempts at characterization of the differences are now 
being made. 

It has been reported by Hirschberg et at. (29) and by Gellhorn (SO) that 
experimental tumors most sensitive to 8-azaguanine have a low concen- 
tration of an enzyme, deaminase, capable of converting 8-azaguanine to 
8-azaxanthine, a noncarcinostatic agent, in contradistinction to those 
tumors not influenced by the compound. Thus, it is suggested that varia- 
tion in response to neoplastic tissues results from their ability to metabo- 
lize 8-azaguanine to an inactive form. In preliminary studies comparing 
deaminase concentrations of Ll210-sensitive and 8-azaguanine-dependent 
cells this does not appear to be the case, since enzyme levels obtained, 
although relatively high, were the same level in both types of cell (81). 
It has been shown also (32) that a combination of 8-azaguanine and 4- 
amino-5-imidazole carboxamide is ineffective in increasing survival of mice 
bearing resistant variants of lymphocytic leukemia L1210, indicating that 
factors other than deaminase content distinguish sensitive from resistant 
leukemic cells. 

Origin of Resistance in Leukemic Cells to Antimetabolites 

It appears extremely likely that the transformations observed in leu- 
kemic cells of the mouse, to resistance or dependence, occur spontaneously 
and rather generally among populations of leukemic cells ; and the role of 
the antimetabolite is merely that of a selective agent (S3) . Increases in 
resistance have been shown to occur in a discrete stepwise fashion resem- 
bling in this respect the development of resistance in bacteria to penicillin 
(34). This observation, along with results obtained by the "fluctuation 
test" and the stability of variant forms in the absence of antagonist, 
would seem to favor the assumption that mutation and selection constitute 
the mechanism by which variant cells arise. It is impossible to determine, 
with these somatic cells, whether the observed transformations are genetic. 
In Escherichia coli, strain K12, a sexually fertile strain, it has been shown 
that the numerous changes to resistance and dependence, using strepto- 
mycin, appear to have arisen by change at a single gene locus (35) . Thus, 
these traits in bacteria behave as if controlled by allelic forms of the same 
gene locus. 

Journal of the National Cancer Institute 


Therapeutic Considerations 

Two different approaches to the chemotherapy of leukemia which appear 
to be of considerable significance are suggested by these studies. The 
first relates to changes in sensitivity to folic analogs of leukemic cells trans- 
formed through the use of purine antagonists; the second relates to a use 
of combinations of antileukemic agents in an attempt to suppress the 
selection of spontaneous mutations to resistance and dependence. 

All three variant sublines of leukemia L1210, dependent upon 8-aza- 
guanine (8-AG-D), resistant to 8-azaguanine (8-AG-R) and resistant to 
6-mercaptopurine (6-M-R) show a striking increase in sensitivity to the 
folic antagonist A-methopterin. This change in response is similar to that 
recorded by Elion and Hitchings (19) in a 6 mercaptopurine-resistant 
strain of Lactobacillus casei which shows a significantly increased require- 
ment for folic acid. 

Table 9 shows this striking difference in response of the three variant 
lines of leukemic cells contrasted with the usual response of the sensitive 
leukemia {23). 

The data of table 10 appear to provide a clear rationale for the use of 
two or more antileukemic agents acting independently. The two most 
effective compounds used in the laboratory, A-methopterin and 8-aza- 
guanine have been shown to act as selective agents in the isolation of 
resistant and dependent mutants. Each also has been shown to act in- 
dependently of the other. Since there appears to be no known method 
for decreasing mutation rates, and it is unlikely that the host can alter 
the process of spontaneous mutation, the approach appears to be an 
attempt to suppress the selection of spontaneously occurring transforma- 
tions to resistance or dependence. The principle of combined therapy 
with two or more agents, acting independently, has been used success- 
fully in infectious diseases, particularly in the treatment of tuberculosis. 
If the frequency of mutation to resistance of a cell, bacterial or cancerous, 
to drug A is 1 X 10~ 6 and a frequency of mutation to drug B is 1 X 10~ 5 , 
only one cell in 10 11 will simultaneously develop both mutations. Thus, 
doubly resistant mutants have a negligible probability of emerging in a 
sensitive tumor or bacterial population in the presence of two or more 
effective agents which exhibit different mechanisms of action. It may be 
seen from the table that these two antimetabolites given singly, in com- 
bination (one followed by the other), at dosage levels below the MTD 
are very effective in increasing survival; when given simultaneously, in 
combination, they exhibit a striking potentiation of effect. Many of 
these mice, though receiving a standard dose (8 X 10 5 cells) of leukemic 
cells live beyond a 90-day period and show no evidence of leukemia. 
These survivors, in all probability, represent cases in which all or most 
leukemic cells were killed and doubly resistant forms completely sup- 
pressed (36). 

The examples discussed here of transformations to resistance and 
dependence involve changes in the cells of the neoplasm. It is conceivable 
that adaptation could occur in cells of the host so that a drug is rendered 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 










o3 ^-s 



d d^ 
-H-H jo 













ft © 




.O •" 





rji CO 







1-1 OS 









53 ^-v 





>• 60 













ft ® 


















lO CO 




•«* cm 

















id cm 




1— 1 1—1 


























> CO 

•gi 1 





0«tf CXJrH^ 





T— 1 1— 1 1— ( 





% g 





lO CO CO tH 

t^ iocx)to 






fc ° 





















'C rt 




•H O +j 

O 03 






pa «h 


,m. CO O 

>» P. <^ 

II « 

O A T3 

<B — < <B 

l 1 I 

M 3 s 

~> o 

■8 S 

o w 

I Si 

> "C3 

~ o 

PI 0J 

C3 m S 

^ 2 B 


I 5 
>> a 

Eh fe 

Journal of the National Cancer Institute 




s c 

8 8 


•<S> «31 

































ci d 

03 03 


* — s> — >. 


v <"• ' 


•H •■-< 


°Q C 5* 

« £ £ 

S dg 

d M a 



(NOO C^TfHCO CDO01>- 'tfCO 

rH CNJ iO ICON -tfOSrH CO!> 

i-HtHCO i-HtHN r-i (M rH (M 



"3 o3 *> 
^73 8 

£ d *- 


02 T 
O Oj 



^.^-.^S^-S ^-S^.^-S^~S ^-V^~S^-S^~S rH CO 

N050 ©<N© OhNO OCOO 

OS<MrHOS OSCN<NOS HlMMffi . . . 

0CXN00O 1> 00 00 OS i> OS IO rH 

rH i-H (N t-I rH rH HrtH _U _jj _M 

HOffliO CO <N h rjH hOhN OS »0 iO 

odt>od© i>os©od od©c©*o i>odos 

HHCO rH<M<N <MrH(N rH<M 

OSOSOS 000000 000000 


uj io io »o a *© *o io o to »o »o 

pji>t-i> pn>i>-t> fli>i>t^ 

o o o 

£^iOiO»0 ^^rHrti ^ tH tJH tH 


cococo cococo cococo 


rH rH rH i— I rH r-f i— I Tf IO CO 



• • OD . • • be 

3 1^06 'o 1^06 

SMa §a-?a 

O«1o6<1 0<Joo<H 



05 bD 

, <C 1 
) 1 00 

fl aa 

S o a h 

" 1 rr\ 1 

■g I bfli- 

§a<? a 






§ o ^3 

^ 2 * 
o * d 

n^ 5 _S 

T3 +i -« 

id a 3 


OS J — 

Q ^!z; 

Vol. 15, No. 3, December 1954 



ineffective. The development of an efficient hepatic detoxication mech- 
anism or an efficient urinary excretion mechanism, for example, may 
render a compound ineffective against neoplastic cells. A known example 
is that by Pollak et al. {87) , who presented evidence which indicates that 
refractoriness of leukemia in mice to potassium arsenite has been con- 
tributed by the host, although other mechanisms appear quite likely in 
this case. 

Recent observations indicate the development of resistance in neo- 
plasms other than leukemia: Sarcoma 180 to 6-mercaptopurine {20), 
the Ehrlich carcinoma to colchicine {38), and the Walker rat sarcoma to 
an acetyl-nitrogen-mustard compound {89) . In this latter case resistance 
is accompanied by a lowered peptidase level. The compound studied 
is believed to act after hydrolysis by a peptidase. 

A recent report {40) is of interest for it indicates the development of 
resistance in normal cells which may prove to be of a distinct advantage 
to the host. Resistance in cells of the intestinal mucosa of the rat to 
aminopterin appears to have developed. The occurrence of marrow 
depression and mucosal damage in patients treated with folic analogs are 
of serious concern to the clinician. These observations suggest the 
possibility of control of those toxic reactions. 


Transformations to resistance and to dependence are found to occur 
rather generally among acute lymphocytic leukemias of the mouse. Two 
types of antimetabolites have been used in the development of these 
variant forms: folic-acid antagonists and purine antagonists. 

Leukemic cells resistant to, or dependent on, folic analogs are inhibited 
by other nonrelated antileukemic agents, but show a characteristic cross- 
resistance (or cross-dependence) to all other 4-amino-substituted folic 
antagonists. Folic acid and citrovorum factor were found not to influence 
the growth of variant cells but both compounds reversed the growth- 
promoting action of A-methopterin in dependent leukemic cells. 

Leukemic cells resistant to, or dependent on, purine antagonist (8-aza- 
guanine and 6-mercaptopurine) show cross-resistance (or cross-depend- 
ence) to all other purine analogs tested, but other nonrelated compounds 
remain carcinostatic. A striking increase in sensitivity to folic analogs 
of all antipurine variants was observed. 

The changes discussed are shown to be stable, irreversible and heritable. 
No reversions to sensitivity, or from one form to another, have occurred 
among the various lines carried in transplant. 

Experimental evidence favors the assumption that the variant cells 
arise by spontaneous mutation, which occurs constantly in populations 
of leukemic cells, the antimetabolites acting as selective agents in the 
isolation and propagation of the variant forms. 

Certain preliminary metabolic studies relating to mechanisms of resist- 
ance and dependence have been given. 

Some considerations of therapeutic interest relating to suppression of 

Journal of the National Cancer Institute 


resistant leukemic cells and the use of altered sensitivity to folic analogs 
have been discussed. 


(1) Law, L. W.: Observations on properties of leukemic cells resistant to folic acid 

antagonists. J. Nat. Cancer Inst. 11: 849-865, 1951. 

(2) : Response of a resistant variant of leukemic cells to an antagonist of 

pteroylglutamic acid. Proc. Soc. Exper. Biol. & Med. 77: 340-344, 1951. 

(3) Law, L. W., and Boyle, P. J.: Development of resistance to folic acid antagonists 

in a transplantable lymphoid leukemia. Proc. Soc. Exper. Biol. & Med. 74: 
599-602, 1950. 

(4) Burchenal, J. H., Babcock, G. M., Beoquist, H. P., and Jukes, T. H.: Pre- 

vention of chemotherapeutic effects of 4-amino-N 10 -methyl-pteroylglutamic 
acid on mouse leukemia by citrovorum factor. Proc. Soc. Exper. Biol. & Med. 
74: 735-737, 1950. 

(5) Nichol, C. A., Zakrzewski, S. F., and Welch, A. D.: Studies on the resistance 

to folic acid analogues in a strain of Streptococcus faecalis. Proc. Soc. Exper. 
Biol. & Med. 83: 272-277, 1953. 

(6) Hutchison, D. J., and Burchenal, J. H.: Observations on the mechanisms of 

resistance to folic acid antagonists. (Abstract.) Proc. Am. Assn. Cancer Res. 
1: 26, 1953. 

(7) Broquist, H. P.: Role of folic acid and citrovorum factor in metabolic processes 

and their relation to cancer. Texas Rep. Biol. & Med. 10: 953-960, 1952. 

(8) Kidder, G. W., Dewey. V. C, and Parks, R. E., Jr.: Growth promotion in 

Tetrahymena by folic acid analogs. Proc. Soc. Exper. Biol. & Med. 78:88-91, 

(9) Nichol, C. A.: Studies of the mechanism of resistance to folic acid antagonists 

by leukemic cells. Cancer Res. 14:522-526, 1954. 

(10) Skipper, H. E.: Progress Report, 1953. Southern Research Institute, Birming- 

ham, Alabama. 

(11) Kieler, J., and Kieler, E.: The effect of A-methopterin on sensitive and 

resistant leukemic cells in vitro. Cancer Res. 14: 428-432, 1954. 

(12) Skipper, H. E., Bennett, L. L., Jr., and Law, L. W.: Effects of A-methopterin 

on formate incorporation into the nucleic acids of susceptible and resistant leu- 
kemic cells. Cancer Res. 12: 677-679, 1952. 

(13) Williams, A. D., Winzler, R. J., and Law, L. W.: The effects of A-methopterin 

on the in vitro incorporation of P 32 (>4 into normal and leukemic mouse tissues. 
Cancer Res. 14: 135-138, 1954. 

(14) Kidder, G. W., Dewey, V. C, Parks, R. E., and Woodside, G. L.: Purine 

metabolism in Tetrahymena and its relation to malignant cells in mice. Science 
109: 511-514, 1949. 

(15) Gellhorn, A., Engelman, M., Shapiro, D., Graff, S., and Gillespie, H.: 

The effect of 5-amino-7-hydroxy-lH-v-triazolo (d) pyrimidine (guanzaolo) on 
a variety of neoplasms in experimental animals. Cancer Res. 10: 170-177, 

(16) Sugiura, K, Hitchings, G. H., Cavalieri, L. F., and Stock, C. C: The effect 

of 8-azaguanine on the growth of carcinoma, sarcoma, osteogenic sarcoma, 
lymphosarcoma and melanoma in animals. Cancer Res. 10: 178-185, 1950. 

(17) Law, L. W.: Studies on the effects of a guanine analog on acute lymphoid 

leukemias of mice. Cancer Res. 10: 186-190, 1950. 

(18) : Resistance in leukemic cells to a guanine analog, 8-azaguanine. Proc. 

Soc. Exper. Biol. & Med. 78: 499-502, 1951. 

(19) Elion, G. B., and Hitchings, G. H.: The mechanism of action of 6-mercapto- 

purine as revealed by microbiological studies. (Abstract.) Proc. Am. Assn. 
Cancer Res. 1: 13-14, 1953. 

Vol. 15, No. 3, December 1954 


proceedings: symposium 

(20) Clarke, D. A., Philips, F. S., Sternberg, S. S., Stock, C. C, Elion, G. B., 

and Hitchings, G. H.: 6-Mercaptopurine: effects in mouse sarcoma 180 and in 
normal animals. Cancer Res. 13: 593-604, 1953. 

(21) Burchenal, J. H., Karnofsky, D. A., Murphy, L., Ellison, R. R., and 

Rhoads, C. P.: Effects of 6-mercaptopurine in man. (Abstract.) Proc. Am. 
Assn. Cancer Res. 1: 7-8, 1953. 

(22) Law, L. W.: Resistance in leukemic cells to an adenine antagonist, 6-mercapto- 

purine. Proc. Soc. Exper. Biol. & Med. 84: 409-412, 1953. 

(23) Law, L. W., Taormina, V., and Boyle, P. J.: Response of acute lymphocytic 

leukemias to the purine antagonist, 6-mercaptopurine. Ann. New York 
Acad. Sc. In press, 1954. 

(24) Heinrich, M. R., Dewey, V. C, Parks, R. E., Jr., and Kidder, G. W.: 

The incorporation of 8-azaguanine into the nucleic acids of Tetrahymena geleii. 
J. Biol. Chem. 197: 199-204, 1952. 

(25) Mitchell, J. H., Jr., Skipper, H. E., and Bennett, L. L., Jr.: Investigations 

of the nucleic acids of viscera and tumor tissue from animals injected with 
radioactive 8-azaguanine. Cancer Res. 10: 647-649, 1950. 

(26) Goldin, A., Greenspan, E. M., and Schoenbach, E. B.: Studies on 

the mechanism of action of chemotherapeutic agents in cancer. IV. Relation- 
ship of guanine and guanylic acid to the action of guanazolo on lymphoid 
tumors in mice and rats. J. Nat. Cancer Inst. 11: 319-338, 1950. 

(27) Gellhorn, A., Hirschberg, E., and Kells, A.: The effect of purines, nucleosides, 

and nucleotides on the carcinostatic action of 8-azaguanine. J. Nat. Cancer 
Inst. 14: 935-939, 1954. 

(28) Bendich, A., Furst, S. S., and Brown, G. B.: On the role of 2,6-diaminopurine 

in the biosynthesis of nucleic acid guanine. J. Biol. Chem. 185: 423-433, 1950. 

(29) Hirschberg, E., Kream, J., and Gellhorn, A.: Enzymatic deamination of 

8-azaguanine in normal and neoplastic tissues. Cancer Res. 12: 524-528, 1952. 

(30) Gellhorn, A.: Laboratory and clinical studies on 8-azaguanine. Cancer 6: 

1030-1033, 1953. 

(31) : Unpublished data. 

(32) Mandel, H. G., and Law, L. W.: The effect of 4-amino-5-imidazole carboxamide 

on the carcinostatic action of 8-azaguanine. Cancer Res. In press, 1954. 

(33) Law, L. W.: Origin of the resistance of leukemic cells to folic acid antagonists. 

Nature 169: 628-629, 1952. 

(34) Demerec, M.: Origin of bacterial resistance to antibiotics. J. Bact. 56: 63-74, 


(35) Newcombe, H. B., and Nyholm, M. H.: The inheritance of streptomycin 

resistance and dependence in crosses of Escherichia coli. Genetics 35: 603-611, 

(36) Law, L. W.: Effects of combinations of antileukemic agents on an acute lym- 

phocytic leukemia of mice. Cancer Res. 12: 871-878, 1952. 

(37) Pollak, M. J., Kirschbaum, A., and Wagner, J.: Refractoriness in the therapy 

of transplanted mouse leukemia. Cancer Res. 13: 39-44, 1953. 

(38) Lettre, H.: Eigenschaftsanderungen von Tumorzellen. Zeitschr. fur 

Krebsforschung 59: 568-580, 1953. 

(89) Danielli, J. F.: The designing of selective drugs. Ciba Foundation Symposium 
on Leukaemia Research. London, J. & A. Churchill, Ltd., pp. 263-274, 1954. 

(40) Vitale, J. J., Zamechek, N., Hagsten, D. M., and Di Georgi, J.: Effect of 
antifolics (aminopterin) on the morphology and metabolism of gastro- 
intestinal mucosa. In press, 1954. 


Dr. Arthur G. Steinberg, Children's Cancer Research Foundation, Children's 
Medical Center, Boston, Mass. 

You see before you a very sad man. Sad because he has lived so long and only now 
has he had the opportunity to meet Dr. Little and his group here at Bar Harbor. 
Sad because now he realizes what he has been missing over the course of the years. 
I hope, somewhat selfishly, that the Jackson Laboratory will have a long, long future 
so that I can catch up on at least some part of what I have missed. 

As I listened to Dr. Law's presentation I became more and more enthusiastic about 
it as long as I could forget that I would be expected to discuss his paper. Then I 
became resentful. I felt that by his excellent presentation of a series of very well 
planned and executed experiments, he was complicating my task immeasurably. 
How can a discussant show his mettle when he agrees with the speaker and has no 
material of his own to present? I must congratulate Dr. Law for a beautiful job, 
beautifully presented. Certainly one cannot take serious exception to Dr. Law's 
interpretation of his excellently designed experiments or to his suggestions for applying 
in the clinic the conclusions he has drawn from them. I would, however, like to draw 
some parallels and indicate some possible cautions. 

I wonder if Dr. Law really insists that the " . . . transformations (to resistance) 
are . . . irreversible." The design of his experiments was such that reversion to 
sensitivity could not be detected. It would be surprising if such reversions did not 
occur because, in general, the ability to undergo reverse mutation appears to be the 
rule rather than the exception. 

There are interesting parallels to Dr. Law's work in the studies of mutations to 
resistance to radiation and to antibiotics in Escherichia coli. Witkin selected 50 
mutants which were resistant to the killing action of ultraviolet radiation. Thirty-one 
of these were also resistant to penicillin and sulfathiazole; 8 were resistant to penicillin 
but not to sulfathiazole; 5 were resistant to sulfathiazole but not to penicillin; and 6 
were resistant to neither. The data make it extremely likely that each resistant mu- 
tant was the result of a single mutation. The last 6 mutants show no cross resistance; 
the remainder show cross resistance to at least one of the antibiotics. Szybalski and 
Bryson showed that the presence or absence of cross resistance to antibiotics depended 
to some extent upon the method of selection. For example, selection for resistance 
to viomycin increased the resistance to this drug tenfold over that of the parent strain 
and simultaneously increased the resistance to streptomycin fivefold. However, 
selection of a strain with greater than a thousandfold increase in resistance to strep- 
tomycin did not result in increased resistance to viomycin. Many other similar ex- 
amples may be found in their paper. We must consider on the basis of the fore- 
going examples that cross resistance to antileukemic drugs may be observed in mice if 
sufficient tests are run. 

The experience of clinicians who treat children suffering from acute leukemia 
suggests that cross resistance may not be infrequent in humans. Some patients who 
suffer a relapse after experiencing a remission induced by one of the antifolic com- 
pounds are resistant to ACTH, to cortisone and to the antip urines; others are not. 
Similarly some patients who suffer relapse after a remission induced by ACTH or 
cortisone are resistant to the antifolics and to the antipurines, while others may 
respond to these drugs and so on. 

Some clinicians stop the administration of the antileukemic agent when a remission 
is induced. Such patients may remain in remission for weeks or months before suffer- 
ing a relapse. I emphasize that during the period of remission the patients do not 
receive any antileukemic agents. Nevertheless, some have been found to be re- 
sistant not only to the agent that induced the remission but also to all other available 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 


proceedings: symposium 

Finally, it should be noted that a considerable proportion of children with acute 
leukemia fail to respond to chemotherapy, i.e., are resistant to all antileukemic 

These illustrations indicate that cross resistance to antileukemic agents does occur 
in humans. Furthermore, they raise the possibility that resistance in humans, in 
some cases at least, may not be determined by mutation in the same way that it is 
in mice. 

I have one more point to offer for consideration. In humans leukemia is treated 
in the host of origin, while, in the mouse, leukemia is treated in a host which has ac- 
quired leukemia via the transplantation of leukemic cells. I wonder what bearing 
this may have on the difference in response of human leukemia and mouse leukemia 
to chemotherapy. I should like to hear the comments of a pathologist on this point. 

Introduction to Dr. Wright 

Dr. C. C. Little 

We move on to the next paper of the program which is a summary of patterns of 
mammalian gene action by Dr. Wright. No one that I know of is as well qualified 
to give the last formal paper here. Dr. Wright, I am sure in the opinion of most of 
us both in the scope of his mind and its penetrating power, is the ranking geneticist 
alive today. Dr. Wright's scientific relationship to Dr. Castle and his membership 
on our Board of Trustees seem to bring together the component elements that make 
it particularly happy to have him in this position on the program at the present 
time. Dr. Sewall Wright of the University of Chicago. 

Summary of Patterns of Mammalian 
Gene Action 1 

Sewall Wright, The University of Chicago, 
Chicago, 111. 

The Gene Concept 

There has been so much questioning of the validity of the whole concept 
of the gene in recent years (1) that it seems necessary for anyone who 
proposes to discuss patterns of gene action to begin with a statement of 
his understanding of what a gene is. 

The controversy probably arises from the increasing divergence of two 
primary concepts. One of these has been that of the gene as a physio- 
logical unit, concerned with a single elementary physiological process, a 
unit character. Alleles are supposed to differ only quantitatively. The 
other concept is that of a morphologic unit, a component of the chromo- 
some that is self-duplicating with respect to its specificity, and that is not 
resolvable into two or more self-duplicating components by crossing over 
or chromosome rearrangement. Under this, alleles are alternative forms 
of a gene, due to mutation within its structure, irrespective of the sort of 
change in physiological effect. 

To a considerable extent there is agreement in practice. Morphologic 
alleles usually affect the same character and often in an apparently graded 
fashion. Heterozygotes of two recessive alleles are usually within the 
range of the homozygous recessives, rather than of the dominant type, 
contrary to the case of heterozygotes of most similar nonalleles. 

It has, however, become increasingly apparent that this agreement is 
far from complete. Careful study usually reveals pleio tropic effects on 
single characters that differ qualitatively rather than quantitatively. In 
extreme cases the major effects may seem to have nothing in common 
(e.g. aristopedia and spineless, dumpy-oblique and vortex in Drosophila 
melanogaster) . Heterozygotes reconstitute type in these extreme cases 
and may even do so where the effects are on the same character [e.g. pro- 
tanopia and deuteranopia in man (&)]. On the other hand, a considerable 
number of cases are now known in which genes that are not alleles in the 
morphologic sense, since separable by crossing over, have such inter- 
dependent effects physiologically that heterozygotes (of recessives) do 
not reconstitute the dominant type [pseudoalleles (8)]. 

It has seemed to me, as I think to most geneticists, that the morphologic 
concept must continue to be the primary one. We could not get along 

» Presented at the Symposium on 25 Tears of Progress in Mammalian Genetics and Cancer, Bar Harbor, Maine, 
June 30, 1954. 


Journal of the National Cancer Institute, Vol. 15, No. 3, December 1954 

838 proceedings: symposium on 25 years op 

very well at the level of the organism without such morphologic terms as 
pituitary, adrenal, ovary, etc., even though we learn that each of these is 
heterogeneous in function and that they are physiologically interde- 
pendent. Similarly, whatever are the most convenient blocks in the 
linkage systems must be named for reference irrespective of physiology. 

The physiological concept suffers from the lack of sharp criteria. No 
one would consider that mutations that are far apart in the same chromo- 
some are alleles merely because they affect the same character (though 
they may be alleles of duplicates). Yet position effects due to a rearrange- 
ment that brings heterochromatin close, but not necessarily very close, 
to a locus are sometimes treated as alleles. There seems to be no evidence 
of any change of pattern at the locus itself and thus no mutation there, 
but merely a fluctuating effect of the heterochromatin on its functioning 
(and often on that of its allele in the homologous chromosome). The 
situation is similar with the remarkable phenomena described by McClin- 
tock (4) and by Brink and Nilan (5) in connection with what had pre- 
viously been considered high mutability at certain loci of corn. Their 
demonstration of a hitherto unrecognized sort of agent, that resembles 
the heterochromatin of Drosophila in inhibiting the functioning of genes 
while in their proximity, but which differs, in frequently shifting by 
itself (under certain conditions) from one position in the genome to 
another, removes this phenomenon from the category of gene mutation. 
What had been considered a single mutable gene, has been resolved into 
two elements that require separate designations. 

The genes as morphologic units are, of course, always provisional in 
character. This leads to a logical difficulty from the lack of assurance 
that there are any blocks in the chromosomes that are wholly exempt from 
crossing over or dissociation by rearrangement, short of single nucleotides. 
Ephrussi-Taylor (6) has indeed shown that many exchanges can occur 
between DNA molecules of pneumococcus of molecular weight about 
500,000 which otherwise one would tend to compare to single genes. 
On the other hand, the very appearance of the salivary chromosomes of 
Drosophila and of the prophase chromosomes of many species especially 
in electron micrographs (7) indicates a heterogeneity of structure in 
which there are blocks that are not wholly arbitrary. More significant 
is chemical evidence [Mazia (8)] that the continuity of the chromosomes 
depends on two different types of bond. It has long been known that 
Drosophila chromosomes may be dissolved by enzymatic splitting of 
peptide bonds. This has suggested a continuous protein backbone. It 
has now been shown, however, that these chromosomes can be broken 
into nucleoprotein particles some 4,000 A° long X 200 A° wide by agents 
capable of binding Ca or Mg. ions (e.g. citrate) followed by a medium of 
sufficiently low ionic strength (distilled water) to permit strong electrical 

It is thus becoming probable that the genes correspond to natural 
physical units (nucleoprotein macromolecules) , bound together to form 
chromosomes by divalent ions. Mazia suggests that this may give a 

Journal of the National Cancer Institute 


physical basis for a sharp distinction betweec intragenic phenomena (point 
mutation) and intergenic phenomena (crossing over, rearrangement) . 

Even, however, if the continuity of the chromosome rested on only one 
type of bond throughout its length, it would still be desirable to name 
regions on as natural a basis as possible. It would hardly be satisfactory 
from the physiological standpoint to limit discussion to the effects of 
mutations. There is undoubtedly differentiation along the type chromo- 
some and different specific physiological effects must be attributed to the 

Gene Duplication 

The most important thing that any gene can do is, of course, to dupli- 
cate itself. The most plausible hypothesis with any substantial factual 
basis has come only very recently. Watson and Crick (9) interpret the 
X-ray diffraction pattern of DNA as indicating a two-strand helix in 
which each strand is a polynucleotide with 10 (or 11) nucleotides per gyre, 
phosphate and pentose backbone on the outside, and cross-connections 
between strands, each consisting of a pyrimidine linked by H bonds with 
a particular purine: cytosine with guanine, thymine with adenine, in 
accordance with spatial relations. There is basis for specificity in the 
order within either one of the strands, that of the other being necessarily 
exactly complementary. They suppose that under certain cell conditions 
there is untwisting and separation, followed by separate recoiling and 
synthesis by each of its complement. This does not explain the relation 
to the polypeptide chains which are always associated with the DNA in 
the duplicating chromosome and which, as noted, seem to give the firmest 
basis for physical continuity within the nucleoprotein macromolecule, but 
is not inconsistent with there being such a relation. 


Among the most important contributions of the genetics of mammals 
and birds are those from the study of antigens, especially of red-blood 
corpuscles. Each antigenic property requires in general only a single 
gene for its determination, and the red cells of an individual, exhibit, in 
general, all of the antigens inherited from both parents without dominance 
or interaction. As noted by Haldane (10), we seem closer to the one-to-one 
relation of preformation here than in any other known character, suggest- 
ing an unusually direct relation between gene and character. 

On the other hand, no other class of loci is known which exhibits such 
complex systems of multiple alleles, each determining a different array of 
antigenic properties. The most complex known seems to be the B locus 
of cattle [Stormont et al. (11)] in which some 80 alleles determine various 
patterns of 21 antigenic properties, each demonstrable by its appropriate 
test serum. Similar cases are known in many other organisms including: 
man (12), fowl (18), and duck (14, 15). 

Those who make physiological considerations primary in the definition 
of the gene would split these loci up into at least as many loci as there are 
sets of antigenic properties that are not alternative in occurrence. Thus 

Vol. 15, No. 3, December 1954 
316263—54 40 


proceedings: SYMPOSIUM ON 25 YEARS of 

M and N in man seem to be strictly alternative and so are S and s but the 
combinations MS, Ms, NS and Ns all occur and are taken to indicate at 
least two loci, even though no crossing-over has been detected. The 
frequencies of these four types in populations are often, however, very far 
from those of random combination. It was shown many years ago by 
Robbins (16) that any initial deviation from random combination falls 
off in each generation of random mating by the mean of the recombination 
percentages in oogenesis and spermatogenesis. The observed deviation 
from random combination thus implies very close linkage, very recent 
origin from mixtures of populations of different origin, or very strong 
local selection for certain combinations. In the case of the II series in 
man in which subdivision into at least C, D and E components has been 
suggested (17) the frequencies of combinations are almost as far as possible 
from random throughout Western Europe. In Czechoslovakia, for ex- 
ample, there is 39.6 percent CDe, 16.9 percent cDE, 40.0 percent cde, 
each two steps from each of the others, and only 3.5 percent of all other 
combinations. The situation is closely similar in England (18). Even 
if this population traced to mixture of three populations absolutely hom- 
allelic in CDe, cDE and cde, respectively, as recently as 300 generations 
ago, which is very improbable in view of the blood groups in the rest of the 
world, crossing-over must be supposed to be less than 0.01 percent per 
generation to account for the rarity of the supposed crossovers. If, 
instead, there is equilibrium between crossing-over and a possible but 
unknown selection against the recombinants, crossing-over must be less 
than 10 percent of the selective differential. 

The hypothesis of multiple loci, moreover, does not account for the many 
cases of asymmetrical occurrence of combinations. In the human ABO 
series, the allele Ai determines two antigenic properties of which one 
never seems to occur except in association with one that occurs by itself 
in A 2 . There are many such cases in cattle. Thus properties B and G 
are found frequently in all of the possible combinations (BG, Bg, bG, 
bg) suggesting separate loci but K, which is very common in BGK seems 
never to be found otherwise. 

It is, of course, very probable that some supposedly single loci will later 
be shown to be compound, in view of the frequency of apparent gene 
duplication indicated by the known cases of pseudoalleles. The criteria 
for splitting loci must, however, be separation by crossing-over or rear- 
rangement, not quality of effect. 

The simplest hypothesis is that the antigenic properties reflect overlap- 
ping aspects of a single varying macromolecular pattern, imposed on the 
antigenic substrate (mucoid carbohydrate in some cases at least) by a 
corresponding pattern in the gene (19, 12, 15, and 11). 

It should be added that even in this class of characters the relation of 
gene to character is not always one to one. Irwin and his associates have 
shown, especially in hybrids of pigeons and doves, that some of the 
antigens which one would expect to find inherited from the parents are 
absent and in their place are new hybrid antigens. 

Journal of the National Cancer Institute 


Reaction Against Grafts 

Mouse genetics has contributed especially to the understanding of 
another aspect of gene-controlled specificity, viz., the reaction against 
transplanted tissues. The subject had two principal roots. Leo Loeb 
(20) } who began its study in the 1890's, found that success of a graft was 
correlated with closeness of relationship: 100 percent success in the case 
of autotransplants, a usually hostile reaction and little success between 
Utter mates (syngenesio transplants) , a more violent reaction between 
unrelated members of the species (homoio transplants) . He considered 
that he was dealing with the heredity of protoplasmic individuality but 
arrived at no detailed genetic interpretation. Little and Tyzzer (21) 
obtained definite Mendelian results which indicated that susceptibility 
depended on the simultaneous presence of a considerable number of 
dominant genes just as the agouti color of the wild mouse depends on the 
simultaneous presence of numerous dominant color factors. This first 
formulation implied that susceptibility itself was the significant character 
that was involved, but later Little and Strong (22) combined the Mendelian 
description with the concept of hereditary specificity. The revised theory 
was that each pertinent gene determines something that tends to induce a 
hostile reaction in a host lacking it. This principle was found to apply to 
the success or failure of transplants of normal tissues as well as of tumors 
[Little and Johnson (23); Loeb and Wright (24)]. Gorer (25) clinched 
this viewpoint years later by showing that one of the genes (H-2) con- 
cerned in resistance to tumor grafts was the same as one that determined a 
red-cell antigen. 

Rejection or acceptance of a tumor graft has, however, turned out to be 
a much less reliable indicator of the presence or absence of "histocom- 
patability" genes than serologic reactions of red cells. If there were no 
complications, the number of such genes (x) in a donor strain, absent in 
the host, could be estimated by finding the proportion of susceptibles in 
F 2 , or a backcross, and equating to (3/4) x or (1/2) x , respectively. It soon 
became apparent, however, that different tumors from the same strain or 
even the same mouse, might give very different results on the same F 2 or 
backcross individuals. One of the most thorough studies of this point 
was that of Cloudman (26) who used 8 tumors from the same closely 
inbred strain (Bagg albinos) in tests of Fi, F 2 and backcrosses to another 
such strain (dilute browns) . His application of the above formulae gave 
results that indicated that one of the tumors had about 12 genes foreign 
to the dilute browns, while the others indicated only 4, 2 or 1. The results 
were consistent for each tumor, except for one that shifted from two to 
one apparent factor in the course of the investigation. It seems clear 
that something tends to happen in the tumor cell that prevents most of the 
antigens from inducing a sufficiently hostile reaction to lead to rejection. 
This could be either loss of the genes or overriding of their effects. Somatic 
mutation is not a plausible answer where so many homozygous dominant 
genes are involved. 

There are various possibilities with respect to the number of genes 

Vol. 15, No. 3, December 1954 



affected. Cloudman's calculations were based on the assumption that 
genes are either wholly effective or wholly ineffective. Another possi- 
bility is that all are equally reduced in the probability of preventing 
acceptance. Finally, penetrance may be reduced unequally. The last 
seems to have been the case. Snell and Higgins {27) have shown that 
there is one gene (H-2 dk ) that is much the strongest and was the one that 
remained effective in the Cloudman tumors in which the apparent number 
was reduced to one. 

The contrast between the expectations under different hypotheses can 
be illustrated by considering one of Cloudman's experiments. He grafted 
the same two tumors (II, V) on each of 105 F 2 mice. The observed 
pattern of acceptances (+) and rejections (— ) was as follows: Expecta- 
tions are given under three hypotheses. 


V=AB 11= V= 

11= V= 




11= AC A (BC) 

A (BCD) 




44. 2 49. 6 

46. 8 




14. 8 7. 9 





14. 8 7. 9 





31. 2 39. 6 






The percentage of acceptances was nearly the same for both tumors 
(54.7% and 52.6%), or approximately (3/4) 2 = 56.25 percent, implying 
only two dominant factors if penetrance is an all-or-none matter. Under 
this hypothesis, one factor must clearly be in common and the other 
different to account for the strong but not perfect correlation. The 
observed numbers do not, however, agree at all well with expectation on 
this basis (probability from x 2 less than 0.01). If however, one factor 
(A, actually SnelTs H-2 dk ) has 100 percent penetrance, and two or three 
others are assumed to have the same lower degree, and the rest none, we 
obtain expectations that agree almost perfectly (B, C both with 58 
percent penetrance) or sufficiently well (B, C, D all with 40 percent 
penetrance, probability 0.20). 

The mode of calculation in the former case is shown below. 




Frequency + + 
27/64 1 

9/64 p 2 
9/64 p2 
3/64 p* 



+ - 
or — + 

2p (1-p) 

2p a-p) 
P 2 (1-P 2 ) 

(1-p) 2 
(1-P) 2 
(1-P 2 ) 2 


Probability arrays of 

No hostile reaction from 

A, B, or C. 
(pC+ + (1-P) c-)» 
(p B++ (1-p) B-)* 
[p> (B+O) + (1-P 2 ) 

Hostile reaction from A 

1 J2 




All ABC mice are expected to accept both tumors. It is assumed that 
C fails to produce a hostile reaction in cc mice in the proportion #, but 

Journal of the National Cancer Institute 


does so in the proportion (1— p) in each tumor, independently of the 
other. The same array of probabilities is assumed for the action of B 
in bb mice. If both B and C are absent in a mouse, the chance of failure 
of penetrance of both is p 2 and of penetrance of at least one (1— p 2 ). 

The equation — - -| — —~- + -rrr = tttf from the cases of double accept- 
64 64 o4 105 

ance leads to the solution p = .418 indicating about 58 percent penetrance 

of either B or C by itself. 

The recognition that penetrance is often incomplete at the level of the 

observed character (rejection or acceptance) has opened the way to studies 

of the physiological factors involved in penetrance, which are being 

actively pursued here under the leadership of Dr. Snell. 

Elementary Metabolic Processes, Enzymes 

There is another class of characters which approaches the antigens in 
the apparent directness of its determination by genes. This is the class 
of elementary metabolic processes, and back of this presumably (and in 
some cases demonstrably) the specific enzymes. A distinction must of 
course be made between those cases in which the molecular pattern of 
the enzyme is altered by an allelic change and those in which enzyme 
activity is merely modified by gene-controlled changes in the conditions 
in the cell. 

Mammalian genetics made the earliest contribution to this field in the 
work of Garrod {28) on "Inborn Errors of Metabolism" in man. Many 
other cases have been studied since. Dr. Sawin of this Laboratory, for 
example, in association with Glick {29) demonstrated that presence or 
absence of atropinesterase in the blood of rabbits depends on a single 
locus. Dr. Chase has described here differences in insulinase in strains 
of mice, but in this case the mode of inheritance seems to be complex. 

The metabolic processes of mammals are difficult to isolate. For a 
systematic analysis of the genetics of metabolic chains we must turn to 
the studies of micro-organisms, initiated by those of Beadle and Tat um 
on Neurospora. These have borne out, with some qualifications, the 
concept of a one-to-one relation between gene and elementary metabolic 
step, on a grand scale. 

The usual one-to-one relation of genes to antigens and enzymes has 
suggested that in these cases the genes either determine directly the pat- 
tern of synthesis of these macromolecules or at least impose a specific 
steric pattern on generalized polysaccharide or protein molecules, in 
either case in direct relation to their own patterns as templates. 

The physiologist traces the properties of cells to the array of specific 
enzymes that they contain and the stimuli which they receive from their 
cellular and humoral environments. Differences in cellular environment 
may trace in part to differences in external environment, but depend 
primarily on the products of other cells and hence trace to enzymes and 
antigens produced in these cells. 

Vol. 15, No. 3, December 1954 


proceedings: SYMPOSIUM ON 25 YEARS of 

Such physiological considerations, as well as those of degrees of genetic 
complexity, suggest that primary action of a gene (including self-duplica- 
tion) may be wholly restricted to imposing a reflection of its own specific 
pattern on various sorts of macromolecules. 

Secondary Characters 

Text-figure 1 is intended to bring out diagrammatically these ideas on 
the relation of genes to characters of various sorts (30-32). At each 
level, physiology is registered in structure which gives a firm basis for 
physiology or behavior at the next higher level. There is no fundamental 
difference between the relation of genes to hereditary extraorganic struc- 
ture (such as that of the webs characteristic of each species of spider), 
and to organic structure. Both are indirect. 






Gene Duplication 

r Homeostasis vs. Disease 






N ) 

\ t 




Mocromolecular Pattern N a 



Text-figure 1. — Diagram of relations of genome and external environment to 
observed characters at various levels of organization. 

Under this viewpoint, we leave the field of gene action (strictly speaking) 
with the determination of macromolecular pattern, and pass into the fields 
of physiology and behavior. Nevertheless, it is convenient to consider 
the types of relation between gene and character at these higher levels, 
skipping over the intermediate physiology. It is not always appreciated 
that the quantitative study of all possible combinations of the genes that 
affect a particular character provides a remarkably delicate technique 
of analysis because of the absence of disturbances due to experimental 
intervention. It is, however, a method that must be associated with 
physiological experiment as far as practicable for adequate interpretation. 

Journal of the National Cancer Institute 


Cellular Characters 

The coat color of mammals is a character at the cellular level since it 
concerns the quality and quantity of certain intracellular products, pig- 
ment granules. In the guinea pig, combinations of alleles at 10 loci 
theoretically make possible more than a million genotypes. What are 
probably the most significant have been studied quantitatively (83). 
Nonadditive interactions are encountered in nearly all cases. The situa- 
tion is similar in the mouse but with an even greater number of possible 
genotypes from known loci. 

Some of the color genes of mammals have been shown to be related 
directly or indirectly to enzyme differences. There are reaction chains, 
somewhat like the metabolic chains worked out in Neurospora. The situa- 
tion is, however, more complex in accordance with the greater complexity 
of the mammal. The genes that affect pigment production in a particular 
cell are in large part those of its own nucleus, but also include those of 
neighboring cells (responsible for tract specificity in coat patterns) and in 
some cases of remote endocrine cells. Migration, differentiation and 
differential viability play roles that require analysis. Thus even here we 
pass beyond the strictly cellular level to that of the organism as a whole. 

The frequency of pleiotropic effects of color factors on seemingly unre- 
lated characters indicate that in many cases changes in pigmentation are 
by-products of more fundamental physiological effects of the genes. The 
classical example is Darwin's reference to the correlation between blue 
eyes and deafness in cats. Many heterozygous whites or near whites in 
mammals and birds are lethal when homozygous. Dr. Eussell (3$.), here, 
has made a notable study of a multiple allelic series of this sort in the mouse 
in which dilution or absence of pigment is associated with effects on 
hemoglobin. Another example is the effect of gene A 7 in the mouse in 
producing yellow color and a disturbed fat metabolism in heterozygotes, 
lethality in homozygotes, referred to by Dr. Chase. 

The Problem of Differentiation 

The strong tendency toward persistence of the type of differentiation of 
cells when removed to indifferent sites, or to tissue culture implies differ- 
ences among cells of the same individual that are hereditary in a broad 
sense, at the level of the cell as a reproducing organism. The nature 
of these self -perpetuating patterns that arise regularly from a single geno- 
type in each generation is one of the most obscure problems of biology. 

There are a number of alternative possibilities. Differentiation might 
depend either on persistent changes in the genome or in the cytoplasm. 
Under the former head comes Weismann's hypothesis of nonequational 
mitosis, now generally discredited, except for certain apparently very 
special cases [e.g. germ line determining chromosomes in Sciara, (35)]. A 
second possibility is mutation of a class of diphasic genes, under the con- 
trol of local conditions. Some support can again be found in special cases 
[e.g. the mutable gene miniature gamma in Drosophila virilis, (36)] but 

Vol. 15, No. 3, December 1954 

846 proceedings: symposium on 25 years of 

no evidence has yet been obtained to indicate that this is a general 
mechanism. A third possibility is differential multiplication of genes in 
the qualitatively identical genomes of different tissues, again under 
stimulus of local conditions, for which Huskins found some support in 
plant tissues and Kosswig and Shengiin [sic] (37), and Sengiin (38) in 
the giant somatic chromosomes of different tissues of Chironomus larvae. 
If on the other hand, the basis is in the cytoplasm, the alternative possi- 
bilities of self-duplicating particles (plasmagenes) and of a repertoire of 
self-regulatory stable states of the cells as wholes, must be considered. A 
form of the latter appears in Weiss' (39) hypothesis that there is determina- 
tion of the cell type by the type of molecule that comes to prevail in its 
surface under the influence of local conditions, including especially the 
surface conditions in adjacent cells, and that once established this tends to 
perpetuate itself. He and others have found evidence that the growth of 
the differentiated cell is regulated in some way by the concentration of 
serologic products of its own, or other kinds of cell, in the blood stream. 
The doubling in the size of one kidney after removal of the other is a 
familiar illustration of such regulation. 

Dr. Dunn's studies of differences in cellular morphology in inbred strains 
of mice bear on this subject. In addition to such primary defects as 
rodless retina in one of the strains, she finds an abundance of character- 
istically different pathologic lesions appearing in tissues (kidney, bone, 
parathyroid, pituitary, etc.) of aging mice of the different strains. 

Cancer in Relation to Genetics 

A cancer cell may be looked upon as a cell in which the pattern of 
differentiation has changed to a new self-perpetuating type of such a 
nature that the normal regulation of growth by the rest of the organism 
is upset. If this is true, the papers on carcinogenesis bear directly on the 
subject of the nature of differentiation. The round-table discussion by 
Doctors Furth, Gardner, Hummel and Woolley dealt with the transforma- 
tion of normal endocrine tissue cells into cancerous ones by mere excess 
or defect of hormones from other endocrine glands. Dr. Furth described 
the induction of thyroid tumors by sustained stimulation of the normal 
thyroid by excess TSH from grafted pituitary tumors and conversely the 
induction of pituitary tumors by removal of the thyroid, whether by 
radioiodine or thyroidectomy. Dr. Hummel and Dr. Woolley discussed 
the induction of ovarian and adrenal-cortical tumors, respectively, by 
various means that prevent estrogenic hormones from reaching and reg- 
ulating the pituitary. 

Dr. Woolley has stressed the importance of strain genotype as a factor 
in the susceptibility to such processes. Dr. Dunn described strain differ- 
ences in the types of tumors that appeared spontaneously with especial 
reference to rare types peculiar to a strain of hybrids. Miss Dickie brought 
out especially the importance of both F x hybrids and backcrosses to 
inbred strains in producing genotypes that in some way lead to the appear- 
ance of rare or hitherto unknown types of tumor. 

Journal of the National Cancer Institute 


Dr. Heston investigated the question whether a genotypic difference 
between two strains with high and low incidence of lung cancer, respec- 
tively, acts through local or general physiological channels, by transplant- 
ing lung tissue from the two strains to Fi hybrids (which accept both, 
since the grafts bring in no foreign genes) . The retention of the character- 
istic differences in rates indicated local action. In experiments on the 
induction of lung tumors by dibenzanthracene, he found a linear relation 
of number of independent tumor nodules to dosage, indicating dependence 
on a single random event in each case. He noted that this is consistent 
with the hypothesis of dominant mutation. 

Dr. Law discussed carcinogenesis as a reversible phenomenon in the 
case of transformation of leukemic cells to apparent normality by various 
chemical agents. 


Turning now to morphogenesis, Doctors Gluecksohn-Waelsch and Chase 
went into a considerable number of cases in the mouse in which mutant 
genes bring about consistent and in some cases drastic changes in the 
pattern of embryonic development. They showed how histologic study 
at successive stages might throw light on the nature of the primary effect 
(e.g., Gluecksohn-Waelsch's account of abnormalities of the notochord 
mesoderm with correlative effects on the nervous system and Chase's 
analysis of the chain of events in hairless mice, tracing to failure of the 
connective-tissue sheath of the follicle to form a glassy membrane that 
holds the latter together at a certain stage) . 

Dr. Gluecksohn-Waelsch referred to the value of what Goldschmidt has 
called phenocopies — environmentally induced abnormalities, identical with 
genetic ones, in analysis of the detailed mechanism by which a gene 
produces its phenotypic effect. Considerable caution is necessary. The 
very fact that identical abnormalities may be produced by such diverse 
means (often including several very different environmental agents) 
implies that neither gene nor environmental agent has anything to do 
with the pattern as such. The physiology of development has remarkable 
self-regulatory aspects by which normality, or a close approach, may be 
arrived at in spite of rather serious disturbances on the way. Some aspects 
are more vulnerable than others. Any disturbance of physiology at a 
critical time may start a chain of processes that results in one of the limited 
number of types of mammalian abnormality determined by the points of 
vulnerability. There is no real copying of gene action by an environ- 
mental agent or the reverse, but a mere triggering by each of a chain of 
processes characteristic of the species. The point on which light may be 
thrown is the nature of the initial abnormal physiological condition, but 
even this may be induced in such widely different ways that little light 
may be thrown on the actual primary action of the gene. 

In most cases, no sharp cleavage can be made between genetic and 
environmental determination of an abnormality. Any given combination 
of genotype and environment, both controlled as far as possible, leads 
merely to the occurrence of a certain percentage of the abnormality in 

Vol. 15, No. 3, December 1954 



question [e.g., otocephalic monsters of the guinea pig, Wright and Eaton, 
(40) ; Wright, (£1)\. It is convenient to speak of a probability distribution 
of physiological states, due to uncontrolled factors, that is cut by the 
threshold for the abnormality in such a way that the area above it is the 
observed percentage. Timofeeff-Ressovsky's term penetrance is a con- 
venient one for this percentage. We have already used the concept in 
connection with the percentage of resistance to grafts of a tumor that 
carries a foreign antigen. 

Dr. Green has made use of this concept in describing the genetics of 
the number of presacral vertebrae in mice. Each of several inbred 
strains had its own characteristic array of frequencies, clearly not due to 
Mendelian segregation, but to a probability distribution with respect to 
nongenetic factors, cut by thresholds. In F 2 and backcrosses, this non- 
genetic variability was of course supplemented by segregation of multiple 
factors. His mathematical analysis was along a line found useful in de- 
scribing the genetics of number of digits in guinea pigs (42). 

Dr. Runner varied both genotype (by use of different inbred strains of 
mice) and the environment (by use of different doses of such teratogenic 
agents as cortisone) in a study of the incidence of developmental anomalies. 
There was always residual nongenetic variability and a threshold. His 
results brought out admirably the symmetry in the roles of heredity and 
environment in the genesis of abnormalities. 

Reaction to Infection 

We come now to a class of characters still more remote from primary 
gene action than morphogenesis — the reactions of the organism to its 
external environment. A subclass is the resistance to infection. Gowen 
notes the tremendous value of inbred strains in revealing genotypic 
differences in this complicated sort of character: reaction to tuberculosis 
in strains of guinea pig (43) and to typhoid (Salmonella typhimurium) in 
mice and (Salmonella gallinarum) in fowls. In his studies of mouse 
typhoid, he found an extraordinary variety of underlying characters 
correlated with resistance, morbidity, and death; e.g., ability to maintain 
weight during the course of the disease, size of heart, kidneys or liver, 
blood volume, hematocrit and leukocyte number, ability of the liver cells 
to isolate the disease and allow the remaining normal cells to function in 
glycogen and fat metabolism, and ability of the macrophages to ingest 
and digest the pathogenic bacteria. Differences in humoral elements — 
albumin and globulin — were significant in acquired resistance. 

Dr. Heston analyzed the genetic control of the transmission of the milk 
agent for mammary cancer on the assumption of normal variability and a 
threshold. The percentages of eliminatiou in repeated backcrosses 
indicated that in this case a very small number of loci differentiated the 
strains used, possibly only one. The relative simplicity of strain diff- 
erences in the response to an agent that is almost part of the heredity of 
the mouse is not too surprising. 

Journal of the National Cancer Institute 



The session on genetical control of behavior dealt with a very different 
aspect of the reaction of the organism to its environment. Dr. Richter's 
paper analyzed some of the physiologic and morphologic changes 
that mediate between genotype and the marked behavior differences of 
domestic and wild rats. He finds evidence of extensive shifting in 
physiology and anatomy of the endocrine glands, probably brought about 
by conscious and unconscious selection for fertility and tameness. 

Dr. Scott showed how the complex differences in behavior among widely 
diverse breeds of dogs could be separated into elements and analyzed 
genetically on the hypothesis of multiple, but not necessarily very num- 
erous, factors. 

Dr. Snyder discussed the supposed selection for lower intelligence in 
man, based on the observed negative correlation between I.Q. and family 
size, in a relatively optimistic vein. 

Formal Genetics of the Mouse 

While most of the authors stressed the complexity of the genotypic 
differences underlying characters at all levels and the consequent im- 
portance of inbred strains, the current advantages of the mouse for analysis 
by conventional genetics were also brought out in the impressive list of 
known genes and their arrangement in linkage systems, presented by 
Miss Dickie. Dr. Griff en showed how far study of the patterns of the 
individual chromosomes and their identification with linkage systems 
has gone. 


We may refer finally to Dr. Castle's interesting account of the early 
history of mammalian genetics, from its beginnings in his own laboratory 
at Harvard almost immediately after the rediscovering of Mendelian 
heredity in 1900. Dr. Castle's early studies laid the foundation for the 
great expansion in which the Jackson Laboratory has played the major 
role in the 25 years since its establishment. 

The results of the Conference gave ample demonstration of the wisdom 
of the program of the Jackson Laboratory in supporting a systematic 
attack on the fundamental genetics of mammals at all levels — from antigen 
to behavior — along with its attack on the problem of cancer. 


(1) Goldschmidt, R. : Chromosomes and genes. Cold Spring Harbor Symp. 

Quant. Biol., 16: 1-11, 1951. 
(#) Waller, G. H. M.: Uber die Erblichkeitsverhaltnisse der verschiedenen Arten 

von angeborener Rotgrunblindheit. Zeit. ind. Abst. Ver. 45: 279-333, 1927. 
(5) Lewis, E. B.: Pseudoallelism and gene evolution. Cold Spring Harbor Symp. 

Quant. Biol. 16: 159-174, 1951. 
(4) McClintock, B. : Chromosome organization and genie expression. Cold Spring 

Harbor Symp. Quant. Biol. 16: 13-47, 1951. 

Vol. 15, No. 3, December 1954 



(5) Brink, R. A., and Nilan, R. A.: The relation between light variegated and 

medium variegated pericarp in maize. Genetics 37: 519-544, 1952. 

(6) Ephrussi-Taylor, H.: Genetic aspects of transformation of pneumococci. 

Cold Spring Harbor Symp. Quant. Biol. 16: 445-456, 1951. 

(7) Yazumi, G., and Keiko, I. : Electron microscopy of salivary gland chromosomes. 

J. Hered. 45: 135-142, 1954. 

(8) Mazia, D. : The particulate organization of the chromosome. Proc. Nat. Acad. 

Sc. 40: 521-527, 1954. 

(9) Watson, J. D., and Crick, F. H. C: The structure of DNA. Cold Spring 

Harbor Symp. Quant. Biol. 18: 123-132, 1953. 

(10) Haldane, J. B. S.: In Perspectives in Biochemistry (Needham, J. and Green, 

D. E., eds.). New York and London, Cambridge Univ. Press, 1938, pp. 361. 

(11) Stormont, C, Owen, R. D., and Irwin, M. R. : The B and C systems of bovine 

blood groups. Genetics 36: 134-161, 1951. 

(12) Wiener, A. S., Sonn-Gordon, E. B., and Hardman, L.: Heredity of the Rh 

blood types. VI. Additional family studies with special reference to the theory 
of multiple allelic genes. J. Immunol. 37: 203-214, 1947. 
(IS) Briles, W. E., Irwin, M. R., and McGibbon, W. H.: On multiple alleles 
effecting cellular antigens in the chicken. Genetics 35: 633-652, 1950. 

(14) McGibbon, W. H.: Further division of contrasting antigens in species hybrids 

in ducks. Genetics 30: 252-265, 1945. 

(15) Irwin, M. R.: Immunogenetics. Advances in Genetics 1: 133-159, 1947. 

(16) Robbins, R. B.: Applications of mathematics to breeding problem: II. Genetics 

3: 73-92, 1918. 

(17) Fisher, R. A.: Population genetics. Proc. Roy. Soc, s.B, 141: 510-523, 1953. 

(18) Race, R. R., Mourant, A. E., Lawler, S. D., and Sanger, R. : The Rh chromo- 

some frequencies in England. Blood 3: 689-695, 1948. 

(19) Irwin, M. R., and Cole, L. J. : Immunogenetic studies of species and of species 

hybrids in doves and the separation of species-specific substances in the back- 
cross. J. Exper. Zool. 73: 85-108, 1936. 

(20) Loeb, L.: Transplantation and individuality. Biol. Bull. 40: 143-180, 1921. 

(21) Little, C. C, and Tyzzer, E. E.: Further experimental studies on the inherit- 

ance of susceptibility to a transplantable tumor, carcinoma (JWA) of the 
Japanese waltzing mouse. J. M. Res. 33: 393-453, 1916. 

(22) Little, C. C, and Strong, L. C: Genetic studies on the transplantation of two 

adenocarcinomata. J. Exper. Zool. 41: 53-114, 1924. 

(23) Little, C. C, and Johnson, B. W.: The inheritance of susceptibility to implants 

of splenic tissue in mice. 1. Japanese waltzing mice, albinos and their Fj gen- 
eration hybrids. Proc. Soc. Exper. Biol. & Med. 19: 163, 1922. 

(24) Loeb, L., and Wright, S.: Transplantation and individuality differentials in 

inbred families of guinea pigs. Am. J. Path. 3: 251-283, 1927. 

(25) Gorer, P. A.: The genetic and antigenic basis of tumour transplantation. J. 

Path. & Biol. 44: 691-697, 1937. 

(26) Cloudman, A. M.: A comparative study of the transplantability of eight mam- 

mary gland tumors arising in inbred mice. Am. J. Cancer 16: 568-630, 1932. 
Snell, G. D., and Higgins, G. F.: Alleles at the histocompatibility locus in the 

mouse as determined by tumor transplantation. Genetics 36: 306-310, 1951. 
Garrod, A. E.: Inborn errors of metabolism. Lancet Jan. 4, 11, 18, 25, 1908. 
Sawin, P. B., and Glick, D.: Atropinesterase, a genetically determined enzyme 

in the rabbit. Proc. Nat. Acad. Sc. 29: 55-59, 1943. 

(30) Wright, S.: Physiological and evolutionary theories of dominance. Am. Nat. 

68: 24-33, 1934. 

(31) : The physiology of the gene. Physiol. Rev. 21: 487-527, 1941. 

(32) : Genes as physiological agents: general considerations. Am. Nat. 79: 

289-303, 1945. 

(83) : Estimates of the amounts of melanin in the hair of diverse genotypes 

of the guinea pig from transformation of empirical grades. Genetics 34: 245- 
251, 1949. 

Journal of the National Cancer Institute 



(34) Russell, E. S.: Analysis of pleiotropism at the W-locus in the mouse: relation- 

ship between the effects of W and W J substitution on hair pigmentation and on 
erythrocytes. Genetics 34. 708-723, 1949. 

(35) Metz, C. W.: Chromosome behavior inheritance and sex determination in 

Sciara. Amer. Nat. 72: 485-520, 1938. 

(36) Demerec, M.: Unstable genes in Drosophila. Cold Spring Harbor Symp. 

Quant. Biol. 9: 145-149, 1941. 

(37) Kosswig, C, and Shengun, A.: Intraindividual variability of chromosome IV of 

Chironomus. J. Hered. 38: 235-239, 1947. 

(38) Sengun, A.: Variability of the banding patterns of giant chromosomes. J. 

Hered. 45: 119-122, 1954. 

(39) Weiss, P.: Some introductory remarks on the cellular basis of differentiation. 

J. Embryol. & Exper. Morphol. 1: 181-211, 1953. 

(40) Wright, S., and Eaton, O. N.: Factors which determine otocephaly in guinea 

pigs. J. Agr. Res. 26: 161-182, 1923. 

(41) Wright, S.: On the genetics of subnormal development of the head (otocephaly) 

in the guinea pig. Genetics 19: 471-505, 1934. 

(42) : The results of crosses between inbred strains of guinea pigs, differing in 

number of digits. Genetics 19: 537-551, 1934. 

(43) Wright, S., and Lewis, P. A.: Factors in the resistance of guinea pig to tuber- 

culosis with especial reference to inbreeding and heredity. Am. Nat. 55: 
20-50, 1921. 

Vol. 15, No. 3, December 1954 

f \ 


Summer Research Awards for College Faculty Members 

The Lalor Foundation recently announced a new program for 1955 to include 20 
summer or interim awards to college and university faculty members for study and 
research in which chemistry or physics is used to attack problems in any of the bio- 
logical sciences. Each award will normally not exceed $900 to single men and women 
and $1,100 to married persons, but is subject to circumstances. The place of work 
may be at the faculty member's own institution or elsewhere, as may fit the best 
interests of the program. 

It is the hope of the Foundation that not only significant research, but also more 
dynamic teaching of science may result from this new program and that younger 
faculty members may find opportunity by this means to advance in their profession. 

Also, the Foundation continues with its eighth year of underwriting of awards of 
postdoctoral summer fellowships administered by the Marine Biological Laboratory 
at Woods Hole, Massachusetts. 

The Foundation is discontinuing its previous program of full-year predoctoral and 
postdoctoral fellowship awards. 

Inquiries respecting the new faculty summer awards should be directed to C. L. 
Burdick, Director of the Lalor Foundation, 4400 Lancaster Pike, Wilmington 5, 
Delaware. Applications for these awards must be filed before January 15, 1955. 

Inquiries regarding Marine Biological Laboratory postdoctoral summer awards 
should be directed to the Marine Biological Laboratory, Woods Hole, Massachusetts. 
The final filing date for these applications is February 1, 1955. 




Date Due 
Returned Due 


■ " IN S' | |tfW» 195 9 

DEC 14 '59 M DEC 1 r ■■;« 

frwtT PEa *-- a Hi l 

jfcp 5 - i frl-jfl 

MRi g>M tip 5 '<* 

JAN 1 9 '62 I 

v ' r '■ • 




C? it. 


of the National Cancer Institute 

Public Health Service National Institutes of Health