412 



NATURE 



{Sept. I, 1887 



The second diagram refers to a similar set of 4380 ther- 

 mometric observations (i) of the 1 aximum, (2) of the minimum 

 temperature on successive days^ from January i, 1873. 



In regard to the first diagram the asymmetry is obvious. I have 

 tested the conclusion in the usual way. For instance, the total 

 of 4857 observations was composed of seven batches of a little 

 less than two years each. Precisely the same asymmetry, in 

 varying degrees, is displayed by each of these batches. The 

 asymmetry is of course obvious to the eye in the diagram, but 

 various numerical tests may be proposed. For instance, we may 

 compare (i) the position of the mean value (in this case 29'9i) 

 between the extreme values, (2) the relative positions of the 

 maximum ordinate and the mean ordinate, (3) the comparative 

 magnitudes of the "mean errors" to the right and the left 

 of the mean ordinate. They all yield a result in the same 

 direction. 



I should be very glad if any of your readers could confirm (or 

 correct) these results by those of more extended observations, or 

 by results taken from other districts. That something of this 

 kind should be displayed where, as here, we are dealing with 

 a one-ended phenomenon — i.e. with one in which unlimited 

 variation was conceivable in one direction but not in the other — 

 seems to me in itself reasonable. But I was certainly surprised 

 to find it so marked, considering how small is the fluctuation in 

 relation to the actual magnitude of the variable phenomenon. 



Fig. 2. 



It seems to suggest that the common theoretic assumption of a 

 sort of fixed mean or type which is swayed about by a large 

 number of equal and opposite independent disturbing causes, 

 does not hold good in this case. 



As regards the second diagram, the two curves are (especially 

 that of the minima) tolerably symmetrical, but they depart 

 widely from anything approaching to Quetelet's supposed fixed 

 type. 



Anyone looking at the curve of maxima would say at once 

 that it mingled the results of two distinct means (in Quetelet's 

 phrase), as if we were to group together the observed statures of 

 a great many Scotchmen and Frenchmen. That we are ming- 

 ling results of distinct means seems true enough, but not of 

 two such, and I cannot account for the two peaks in the curve. 

 What I should have expected would have been something of this 

 kind : Each day has its own appropriate mean maximum (subject 

 to the usual fluctuation), and these mean maxima are themselves 

 grouped about thdr mean, hence the true mean of all ought to 

 be decidedly the commonest result, i.e. the curve should have a 

 single vertex. 



The facts are quite otherwise. The depression towards the 



In this case, as the lengths of the successive ordinates from the original 

 data were very irregular. 1 have smoothed the curve out by taking the mean 

 of three .successive heights. For instance, to take the actual figures, tlie 

 number of occasions on vv-hich the ma.\ima vi ere 58% 59°, and 60°, were 

 respectively 108, 99, and 124; I have assigned the number 110 to 59°, and 

 so on. 



centre is far too deep to be accidental, and the final mean {i.e. 

 about 57°) is very far from being the commonest value. 



Somewhat similar remarks may be made about the curve of 

 minima. There is some evidence (though not conclusive) of a 

 depression towards the centre in this case also, and the curve is 

 very fairly symmetrical. But the true mean of all the minima 

 cannot claim any numerical preponderance over any other value 

 between 32° and 52°. 



I am far too deeply conscious of the nuinerous pitfalls which 

 lurk about the statistician's path to offer these results with any 

 great confidence. But considering how large is the number of 

 observations included, it certainly seems to me that they call for 

 some explanation. There may of course be some blunder in the 

 calculations, but I have done my best to guard against this. What 

 I trust is that these results may be the means of calling forth 

 some discussion by practised experts in this branch of statistical 

 inquiry, which may serve to confirm or correct my results, and 

 in the former case to offer some explanation of the causes of the 

 phenomena. Very likely this practical inquiry has been already 

 undertaken elsewhere, but the statistics of meteorology are so 

 vastly extensive that it is impossible for any but a professional 

 student of the subject to be acquainted with what goes on 

 in it. J. Venn. 



Cambridge. 



The Sense of Smell in Dogs. 



Will Mr. Russell (whose letter in Nature of August 4 I 

 have just read) be so good as to make another experiment with 

 his pug bitch? lie says that she had been " taught to hunt " 

 for biscuit ; probably she was also enjoined to ^^ find it," or 

 something similar, when she came into the room. Can he 

 manage to try her powers without awakening her expectation ? 



I ask it because it seems to me that in this case (and many 

 others) we have something different to observe than mere quick- 

 ness or keenness of sense, and something well worthy of obser- 

 vation ; namely, exclusive direction of the attention of a sense — 

 if I may so term it. 



We may note this mysterious power in ourselves to a certain 

 extent. In the case of a dog or bird, or any other in which there 

 is little brain work going on to cause distraction, it may be much 

 greater, and account for many wonderful things. It may be said 

 that this is trying to explain the unknown by the even less 

 kiiowri ; nevertheless, by gathering together many and varied 

 instances of the action of any power some light mtist be thrown 

 upon it. The mesmerizer seems to deal with this one when he 

 closes all avenues to the senses of his subject except the one he 

 wishes to keep open. 



The sense of hearing in some birds seems as wonderful and 

 discriminating as that of smell in dogs. I have watched with 

 astonishment a thrush listening for worms — as their manner is — 

 and very evidently hearing them too, within two yards of a 

 noisy lawn-mower on the other side of a small hedge of roses. 

 Probably the worms came nearer to the surface in consequence 

 of the vibration caused by the machine — they are said to do so 

 — but that the thrush heard and did not see them was evident. 

 Robins appear to be able to distinguish the voices of their own 

 offspring and parents from a number of others, and at a great 

 distance. I say appcai; for in such a case one cannot be quite 

 sure, still less can one give all the small details of long-continued 

 observation that make up the evidence in favour of it. 



All these cases have a common and mysterious element. It 

 is as if a window were opened in one direction and all others 

 closed ; or a chord set vibrating that answers, as a struck glass 

 answers, only to one note ; or as if all the available energy were 

 directed along one narrow path. At any rate there is something 

 more than mere keenness of sense. J. M. H. 



Sidmouth. 



Electricity of Contact of Gases with Liquids. 



Will you allow me to ask Mr. Enright (Nature, p. 365) 

 how he proved that the " charge of the escaping hydrogen was 

 positive " or negative, as the case may be ? That the escaping 

 spray was electrified by friction, after the manner of the steam 

 spray in Armstrong's old hydro-electric machine, is a natural 

 explanation of these capricious effects ; but that gas should be 

 thus electrified, and that this electrification should have any 

 relation whatever to the subject of "atomic charge," are 

 propositions which strike one as improbable. 



Oliver J. Lodge. 



