APRIL 29, 1897 | 
NATURE 
605 
merely a collection of these spheres in a very high state 
of condensation. To illustrate this tendency to condense, 
the author compares it with the tendency of the sun and 
stars to cool and contract, and eventually to form bodies 
like the moon. This theory of the constitution of mattet, 
we are told, explains all the natural phenomena of light, 
heat, electricity and magnetism, without a single con- 
tradiction. 
Having tried to upset the existing theories, and having 
told us that this new theory will explain practically 
everything, the author, to our surprise, fails completely 
to put forward convincing proofs in support of its appli- 
cation to electricity and magnetism. A good example 
of the class of explanation with which the pamphlet 
abounds is to be found on page 48, where the loss of 
electricity from conductors in damp weather is alluded 
to. A positively charged body is supposed to be sur- 
rounded by layers of negative ether spheres—that is, by 
spheres having a larger radius than the mean ether 
sphere. These negative ether spheres are in a high 
state of tension, and when a water molecule comes into 
the space which they occupy it relieves this tension, and 
so partly discharges the conductor. If we accept this 
explanation, there is absolutely nothing to prevent us 
supposing that small particles could discharge a con- 
ductor without touching it, or without being connected 
to it by any other material except the ether, as the 
author supposes that the layers of negative ether spheres, 
above alluded to, extend to finite distances from the 
conductor. 
Experience, however, will not allow us to accept such 
an explanation at all, for it has been perfectly well 
established that the vapour rising from an electrified 
surface carries with it no charge. In connection with 
the magnetism of the earth we find, on page go, an 
interesting piece of information. We are there told that 
it is only those heavenly bodies which rotate that have 
polarity, and that, corseguently, the moon is non-magnetic! 
It is consoling to learn that the author has suffered 
hitherto so much from hostile critics that he can no 
longer be stung by the suggestion that his philosophy 
is “blank Unsinn.” We Ss “Ap 
Farm and Garden Insects. By Prof. Wm. Somerville. 
Pp. viii + 127. (London: Macmillan and Co., Ltd., 
1897.) 
A USEFUL little ¢ea?-Go0% for beginners, and an excellent 
xeference book for practical farmers. The three parts 
into which the book is divided are judiciously arranged. 
The first gives in a clear and distinct manner the rudi- 
ments of entomology, and forms, therefore, a useful 
introduction to the second part, which describes some 
of the most common insect pests whose ravages cause 
so much loss to the farmer and gardener. This loss may 
be very much modified if the simple precautions and 
remedies contained in the book are adopted. The 
appendix in a few pages gives most useful information 
about mites, ticks, &c. ; not true insects certainly, but 
which, by attacking our domestic animals, and even man 
himself, cause an immense amount of irritation, inflam- 
mation, and consequent loss. Farmers, gardeners, and 
all interested in rural economy will do well to carefully 
study its pages. 
Geology of North-east Durham. By D. Woolacott, B.Sc. 
Pp. vit 84. (Sunderland : Hills and Co., 1897.) 
THIS is an orderly account of the geological charac- 
teristics and history of North-east Durham. It is written 
in language easily understood by readers unacquainted 
with the elements of geological science, and will, there- 
fore, interest a pepular public as well as the student of 
British geology. The diagrams are very coarse ; but this 
is doubtless due to the fact that they were prepared for 
publication in a weekly newspaper, in which the articles 
now reprinted originally appeared. 
NO. 1435, VOL. 55] 
LETTERS TO THE EDITOR. 
(Zhe Editor does not hold himself responsible for opinions ex- 
pressed by hts correspondents Netther can he undertake 
to return, or to correspond with the writers of, rejected 
manuscripts intended for this or any other part of NATURE. 
No notice zs taken of anonymous communications. | 
Rate of Racial Change that accompanies Different 
Degrees of Severity in Selection. 
Ir is well known, in a general way, that better results are 
obtained by breeding from very select specimens than from the 
less select ; but the statement deserves to be expressed with 
greater precision. I will do so here, on the lines laid down in 
“Natural Inheritance ” (Macmillan, 1889), using the constants 
there determined for the stature of British men, including among 
them the coefficient by which female stature may be corrected 
to its equivalent male value, and thereby eliminating all trouble 
due to sexual differences, 
On this basis, it will be shown, by way of illustrating a 
general problem, how much the stature of a breed of British 
men would be raised in each successive generation, under 
different specified degrees of severity in selection; also the 
utmost limits of stature to which they could be raised in the 
several cases, no change of type being supposed to occur in 
the interim. 
Degrees of severity in selection admit of being defined by the 
method of cezfzles (or percentzles) fully described in the above 
book. No ambiguity need arise in interpreting such a state- 
ment, as that a man occupies the ninetieth centesimal grade in 
stature among a population whose mean stature is 68°5 inches, 
and whose individual statures are normally distributed about 
that mean with a quartile of 1°5 inches. Referring to Table 8 
in the book, it is seen that the normal deviation at 90° in a 
series whose quartile is 1, is 1°90; therefore, in the above case, 
its value is 1°9x1°7 inches=3°23 inches. The mean stature 
of the population is 68°50 inches, which has to be added to 
this, making a total of 71°73 inches. Consequently when it is 
said that those persons are selected for parents who occupy the 
grade of go’ in their respective series, the degree of severity in 
the selection has been strictly defined, Similarly in respect to 
any other grade, such as 80° or 70. This method of defining 
severity of selection is applicable to every measurable character, 
and to every form of distribution, skew or other, revealed 
by observation. 
The principle has been fully explained in the above book, by 
which successive generations of the same population are able to 
maintain the same statistical peculiarities notwithstanding the 
“*scatter” of fansilies. It was shown that the sons of parents of 
similar statures form a co-fraternity, whose mean is more 
mediocre than the parental statures. In other words, the mean 
of the co-fraternity vegvesses towards the mean of the race, the 
coefhicient of regression in stature being 2/3. Thus the children 
of parents of grade 90° in stature, deviate on the average no 
more than 2/3 x 3°23 inches, or 2°15 inches, above the mean of 
the race. So much for the first generation of the selected 
parents. 
In the second and subsequent generations, the ‘‘scatter” of 
the co-fraternities has to be considered. The quartile of every 
one of them was shown in the book to be 1°5 inches, conse- 
quently the individuals who occupy the grade 90° in a co- 
fraternity, are I°5 x 1°90, or 2°85 inches taller than the mean of 
the co-fraternity, which itself is 2°15 inches above the mean of 
this race, making a total of 5’00 inches. The mean of their 
offspring, that is of the individuals forming the second gener- 
ation of the selected series, is 2/3 of 5'00 inches, or 3°33 above 
the mean of the rest of the race. 
These results are easily generalised and thrown intoa formula, 
as follows: w=coefficient of regression ; ‘=tabular deviation 
at the specified grade ; g=quartile of race ; g’=quartile of co- 
fraternity; a=/g; B=¢7'. Then the mean deviation of the 
pedigree stock from the mean of the race, in each successive 
generation, is :— 
Ist generation, wa. 
2nd ns ze(zwa +B). 
zw — Ww" 
nth a wo"'a + arr 
When is large, w’ disappears and the limiting value be- 
w 
comes 2 
I =e 
to 2B. 
If w=% as above, the limiting value is equal 
