them was built up of fluctuations. 

June 19, 1923] 
NATURE 
807 

1. Our powers of observing are proportions to 
our familiarity with the objects of study. In all 
Nature, we are, for purposes of observing fluctuations 
and natural selection, sufficiently intimate with only 
one species—our own (see NATURE, January 13, 
p. 50). Either we must derive our evidence from 
the lives and deaths of men, or else our thinking is 
mere guessing. Examining the evidence, we find 
(a) that stringent natural selection is in full swing 
all the world over; (b) that fluctuations, not mutations, 
are chosen ; and (c) that the result is always adaptive 
evolution. If that be the case with men, whom 
naturalists describe as having “ escaped from selec- 
tion,’’ is it not, to say the least, likely to be the case 
with other types ? 
2. Human races never differentiate while there is 
interbreeding, but invariably diverge when separated 
by time and space. So far as clear evidence goes, 
this is true of all natural types. Have we not here 
proof that offspring blend parental differences, and 
therefore that natural racial change is based on 
fluctuations ? 
3. It has been said very truly, ‘‘ The fact that the 
gametes ofthe cross transmit each member of the 
pair pure, is as strong an indication as can be desired 
of the discontinuity between them.’’ The converse 
must be true also; if the members of the cross 
blend, we have proof that the unlikeness between 
Now, all human 
races, and so far as I know all natural races, blend 
when crossed, except in characters linked with sex. 
4. There is massive evidence, not disputed by 
Mendelians, that male is undeveloped female, and 
vice versa. Here we have alternate patency and 
latency, alternate reproduction, not alternate in- 
heritance. Male and female characters, belonging to 
different sets, do not blend; but, presumably, the 
onde characters of the one sex blend with the 
tent characters of the other. Sometimes, as in 
aphides, the patency and latency extends unaltered 
Over many generations. But if, as is alleged, the 
inheritance of sex is Mendelian and segregation 
occurs, how is the prolonged latency of the male 
traits conceivable ? Theoretically males, as well as 
females, should appear in the first parthenogenetic 
generation, after which, since the males cannot 
reproduce, and the females have become pure 
dominants, males should never again appear. 
5. If a mutant crosses with the normal, we have, 
admittedly, in the impure dominant, exactly such 
ny and latency as is found in sexual characters. 
endelians insist that afterwards there is segregation, 
and, therefore, that in following generations pure 
dominants and recessives occur. How, then, does 
it happen that ‘‘ pure ’’ dominants sometimes produce 
recessives, and recessives dominants ? How does it 
happen that purely bred domesticated types (e.g. 
igeons, poultry, and many plants) often ‘“‘ throw 
ack ’’—reproduce the ancestral type which according 
to theory was eliminated perhaps hundreds, perhaps 
even thousands, of generations previously? Is this 
not clear evidence that Nature treats mutations like 
sexual traits, making their reproduction, not their 
inheritance, alternate ? 
6. Crossing often reveals long-lost ancestral traits 
among artificial varieties, but never among men or 
other natural types. Is this not clear proof that while 
man often chooses mutations, Nature selects among 
fluctuations ? But man does not always choose 
mutations. Sometimes, though he cannot easily 
perceive fluctuations except among his own kind, he 
selects them. Thus speed in racehorses is due to 
a high average of excellence in a thousand co- 
ordinated structures. A thousand mutations occur- 
NO. 2798, VOL. 111] 
_ Mendelian word “ gene.” 
ring at once are out of the question. As might be 
expected, (a) racehorses tend, in lack of stringent 
selection, to retrogress with a speed which is pro- 
portionate to the antecedent progression, and (b) the 
offspring of a cross between race and ordinary horses 
blend the parental differences, as in hunters and hacks. 
7. Apparently, then, the crossing of mutants with 
the normal results in alternate reproduction. Yet 
careful artificial breeding produces all the effects 
of alternate inheritance; for in this way undesired 
traits may be rendered almost permanently latent, 
as witness the narrow stripes revealed by the offspring 
of horses crossed with the broad-striped Burchell 
zebra, whence it appears that man never domesticated 
"the horse as such, but began with an animal striped like 
the Somali zebra, the coloration of which he rendered 
latent by selecting mutations. Human mutations— 
idiocy, hare-lip, tumours, and so on—are common; 
but useful human mutations are unknown. Many 
of them, idiocy, for example, become yearly more 
common, and medical men, believing that they 
usually indicate the reappearance of latent ancestral 
traits, hope by preventing procreation to reduce 
their frequency. If, however, Mendelian theory be 
correct, they are new variations, and the position of 
humanity is hopeless. Inthat case the human species 
is ‘‘ ever-mutating ’’; and, since human mutations are 
ever-injurious, there is no scope for selection. And 
yet the odd thing is that there is always selection, and 
the result is always improved adaptation. 
8. I suppose my word “ predisposition ” (potenti- — 
ality, diathesis) corresponds’ somewhat with the 
But I merely follow physio- 
logists and pathologists who, assuming germinal 
predisposition, seek in each case to discover the 
nurture that causes development. If I have specu- 
lated, it is only to suppose, as others have done, that 
perhaps all the cells of the body are alike in nature, 
and differ only through nurture. ‘‘Gene,’’ on the 
other hand, implies a knowledge more profound, or 
an assumption more daring. Itself a discrete unit, 
it is the representative of a unit character (one with 
Mendelian inheritance). Like the “ physiological 
units,’’ ‘‘ gemmules,’’ and “ determinants ’’ of our 
predecessors, it is a brick in the architecture of the 
ermplasm. It is difficult to understand, however ; 
or the multicellular individual, a cell-community, is 
compounded of characters ; and a character may be a 
sub-community (e.g. rose-comb, extra digit, black- 
smith’s muscle), or a quality of the whole com- 
munity (for example, size, shape, colour), or a quality 
of a character (for example, colour of a flower). As 
we have just seen, crossed natural varieties blend 
their characters: have they, then, no unit charac- 
ters, and, therefore, no genes? I understand that 
characters, not their modifications, are represented 
by genes. Thus there is a gene (or genes) for normal 
but not for diseased skin, for skin-colour but not for 
sunburn, for normal but not for blacksmith’s muscles, 
Unfortunately I know of no character that is not a 
modification of its antecedent self. Thus the muscles 
of the athlete were modified by nurture from those 
of the ordinary man, which were modified from those 
of the youth, and so on right back to the germ-cell. 
Which, then, is the character, and which its modifica- 
tions? Obviously all characters are “‘ fluctuations 
due to conditions of environment, to nutrition, 
correlation of organs, and the like. There is no 
indisputable evidence that they can be worked up 
and fixed as a specific character.” But try to con- 
ceive a character which is not a product of some sort 
of nurture! It appears, then, that genes represent 
nothing conceivable, and that evolution is impossible. 
But the attempt to understand genes makes my head 
