712 ; 
be marked “right,’’ but a W.N.W. wind “ wrong,” 
because this type of weather accompanying the two 
is different. So with regard to easterly winds, if, say 
N. is predicted and E.N.E. follows, this would be 
“right,” but an E.S.E. following on the prediction of 
N.E. would be ‘‘ wrong.” 
With regard to strength, the forecasts often make 
anything less than a gale a possibility. 
The temperature forecasts are generally vague, but 
if a rise of temperature occurs when the prediction is 
“cool,” this would be marked ‘‘ wrong,” or, again, if 
moderate temperature is predicted and the true tem- 
perature differs much from the average for the time 
of year. So far as possible I have given a favourable 
interpretation to all the forecasts with the results 
which are here shown in Table I. 
TasLe I.—Comparison of Weather in London in 1913 
with the Forecasts made on the Previous Day. The 
actual Weather is Judged from the Morning Weather 
Chart of the Day. 
Wind Temperature 
= ee 
Right (Wrong, 2 [No fre! Right Wrong] Doubt: |No fore- 
ee ss EE Re ae = 
January... | 18 Gin) = 4 U5}, 110 2 4 
February..| 13 | 11) —| 4 14 | 8 2 4 
March ...| 14 vial eee 6 13 7 4 6 
ASDC cae 15 10 I A ASH | SAO 4 4 
Maye) 2.1( “Ig | ae: 4 a ean 2 4 
Juner.../| 7 6| 2 5 57 2 5 
july S Bex | 20 On 4 G2 Lar 4 4 
August ...| 14 8 | 4 5 15 9 2 5 
September] 12] 12) 2 4 13| 8 5 4 
October... | 13 £34); ad 4 15 10 2 4 
November| 13 CS ee 7 free hich’ 7 5 7 
December | 19 5 3 4 || 14 8 4 4 
Total ... | 181 | 107 | 22 | 55 ‘|| 161 | 109 | 38 55 
Percentage | | 
on No. of | 58 | 34°5|7°5| — B20 36 12 — 
forecasts | } 
So far therefore as the present very limited examina- 
tion goes, the probability of finding the predicted 
weather on the morning of the day succeeding the 
forecast is as follows :— 
Wind Temperature 
Probability for 0°58 0°52 
FE against 0°35 0°36 
4 doubtful 0°07 o'l2 
Taste I].—Change of Type of Weather. 
i Changes of Forecast Beverage 
yPe Right Wrong forecas's 
January... ... 15 im 4 73 
February... 10 8 | 2 80 
March: ..< v7 9 4 | 5 44 
Apel y Sve 0 ites II 7 4 63 
Manus 555. dis 14 10 4 71 
Jue ....> of 6 4 2 66 
UU ee eer ta 12 10 2 83 
FAUIOUSE nes 9 4 5 44 
September ... 13 7 6 54 
October axe 18 15 3 83 
November ... 12 6 6 50 
December. 9 7 2 78 
Motels 2)... 138 03 45 789 
es ae es | re —}| —— —_|——____ —_ 
Percentage . _— 67°4 32°6 — 
NO. 2313, VOL. 92| 
NATURE 
. 
[FEBRUARY 26, 1914 
This is intended to indicate the corréctness of the 
forecast as regards change of weather, and here the 
percentage of success is higher, being 67-5. 
It will be noticed that during the year the type of 
weather changed 138 times. If, therefore, anyone 
with no information whatever had been content with 
saying ‘To-morrow will be like to-day,” he would 
have been wrong 138 times, and right 227 times, and — 
the probability in favour of his prediction being 0-62, — 
and against 0-38. With the help of a barometer and 
ordinary local observation, he might probably improve 
on this somewhat. P 
Comparing these figures with the results obtained — 
by such a fairly complete knowledge of the simul 
taneous conditions in surrounding regions as is 
afforded by telegraph day by day to the Meteorological 
Office, it does not appear that for this latitude and 
country the odds in favour of the latter are large. 
On looking over the immense mass of figures— 
considerably more than 1000 entries—which are used 
in the construction of each Weather Chart, I am 
reminded of a sentence in the late Sir G. (then Mr.) 
Airy’s report to the board of visitors of the Greenwich 
Observatory in 1867, when speaking of the proposed 
increase in the number of meteorological observatories, 
“whether the effect of this movement will be that 
millions of useless observations will be added to the 
millions that already exist, or whether something may 
be expected to result which will lead to meteorological 
theory, I cannot hazard a conjecture.” De Morgan 
quotes this sentence (the last in the report in ques- 
tion) in his ‘‘ Budget of Paradoxes,” and remarks :-— 
“This is a conjecture, and a very obvious one: if 
Mr. Airy would have given 23d. for the chance of a 
meteorological theory formed by masses of observa- 
tions, he would never have said what I have quoted.” 
Personally, I think it extremely improbable that any 
trustworthy weather forecasts for periods so long as 
twenty-four hours will, or can, ever be made for 
regions outside latitudes 30 N. or S. or thereabouts, 
with the exception perhaps of a few places where the 
local conditions are paramount. 
The reasons for this view will be given in a subse- 
quent note. A. MALiock. 
February 17. 
The Darwinian Theory of Atolls. 
In the review of “ Letters and Recollections of Alex- 
ander Agassiz’? (NATURE, January 29)—an article in 
which we are given a picture of an exceptionally in- 
teresting personality drawn by an intimate friend— 
there is a brief paragraph on the formation of coral 
reefs to which I should wish to refer. . 
My friend, Sir Ray Lankester, concludes that 
Agassiz ‘certainly succeeded in showing that the 
views advocated by Darwin and by Dana are not 
capable of universal application, nor, indeed, of general 
validity ”’ (pp. 603-4). 
It would be unreasonable to claim “universal appli-— 
cation,”’ but ‘‘ general validity” is a matter to be deter- 
mined by a comparison of the number and extent of 
the regions in which Darwin’s explanation holds good 
with the number and extent of those in which it does 
not. The results of the Funafuti boring offer, I think, — 
important evidence in regard to the Pacific area. A 
test locality was chosen with the utmost care by un- 
biassed authorities. A bore-hole was drilled, at great 
expense and in spite of many difficulties, to a depth 
of 1100 ft. Only shallow water organisms were 
found in the core. For some reason or other— 
probably because it is more exciting to overturn than 
to. confirm—very little has been said about this 
evidence. We all hope for more borings; but in the 
meantime the only important trial that has been made 
