Feb. 25, 1875J 



NA TURE 



;27 



there is nothing adventitious ; the pipe is a mechanism designed 

 to a precise end which it fulfils ; it speaks but as it must ; there 

 is no selective power, for the hand that fashions it, ordains. 



Hermann Smith 



Periodicity of Rainfall 



In his second letter (Nature, vol. x. p. 263) Governor Rawson 

 makes the following remarks: — "Mr. Meldrum, in his letter 

 (vol. viii. p. 547), writes, that I have 'taken 1S46 and 1871 as 

 middle maxima years \in my first paper I also took 1S48], 

 w/iereas 1849 and 1872 are probably more eorrecl.' Mr. Meldrum 

 is in error as to my having taken 1846 as a middle maximum, 

 as a reference to my former letter will show. ... I demur to 

 the changes to 1849 and 1872 : to the first because, without any 

 sufticient reason, a dry year (48'to in.) is discarded, and a wet 

 year (67 '88 in.) is added ; and to the second, not because it 

 affects my calculations, but because no reason is given." 



In reply, I beg to observe that 1S46 is either a misprint for 

 184S, or that in my manuscript 6 was inadvertently written for 8. 

 This, I submit, is evident from the words immediately following 

 the mistake, namely, "in my first letter, / also took 1848." 



If Mr. Rawson supposes, or if his remarks imply, that I made 

 1849 a middle maximum, to avoid the small rainfall of Barbados 

 in 1S47 (48'lo in.) and at the same time to take advantage 

 of the large fall in 1850 (67S8 in.), in order to make out a 

 favourable case, I beg to say that he is entirely mistaken ; for 

 long before I saw his rainfall returns, I had invariably taken 

 1S49 ^s a middle maximum year. The only instance in which 

 I took 1848 was, as I said, " in my first paper " read before the 

 Meteorological Society of Mauritius on Oct. 10, 1872. In all 

 subsequent papers on the subject, including one read before the 

 Royal Society, 1S49 was taken. Righty or wrongly, therefore, 

 the Barbados rainfall h.as been subjected to exactly the same 

 treatment as that of the British Islands, the Continent of Europe, 

 India, America, &c. 



Assuming a causal connection between sun-spots and rainfall, 

 it seemed to me that the effects, if any, would be most apparent 

 about the times of the turning-points of the sun-spot curve, and 

 that a comparison of the rainfall of each maximum period of 

 three years with that of each minimum period of three years, for 

 a considerable time and space, would be a preliminary test of the 

 hypothesis. The difficulty was to know the exact epochs of 

 maximum and minimum sun-spot frequency, and at the same 

 time the rainfall for equal periods on either side of them. If we 

 had the monthly rainfalls, and knew in what month the maxi- 

 mum and minimum of sun-spots occurred, it would be compara- 

 tively easy to compare the rainfalls for equal times with respect 

 to the epochs. But there was another point to be considered, 

 namely, that a cause requires time to produce its effect. 



According to Prof. Wolf 184S.6 was a maximum epoch; 

 which, I presume, means that the turning-point occurred in 

 August 1 84S ; the figures, however, might mean six-tenths of a 

 year after 1848, or August 1S49. 



Taking August 1S4S lor the maximum epoch, the strict course, 

 in order to place the epoch at the middle of thirty-six months, 

 would be to give the rainfall from the 6th of February 1847, to 

 the 6th of February, 1S50. But this could not be done. It was 

 necessary to choose a zv/iole year as the middle maximum year. 

 And the reason why 1849 was chosen in preference to 1S4S was, 

 that the object being to find whether the periodic changes indi- 

 cated by sun-spots had any effect upon rainfall, and time being 

 required for a cause to produce its full effect, there was a pre- 

 sumption that the maximum rainfall would take place after the 

 maximum of sun-spots, somewhat in the way in which the 

 maximum diurnal temperature occurs, not at noon, but an hour 

 or two after noon. 



For a similar reason 1872 was taken as a middle maximum in 

 preference to 187 r. 



This allow.ance of time for the supposed cause to produce its 

 effect is, though apparently unintentionally, made by Mr. 

 Rawson himself when he adopts 1S44, 1856, 1S60, and 1867 as 

 middle years ; for, according to Wolf, the epochs were 18440, 

 lS56'2, lS6o'2, and iS67'i, that is, if I mistake not, early in 

 each year ; so that nearly two of each of the three years taken 

 come after the epoch, while only one of them precedes it. By 

 taking 1S49, therefore, as a middle maximum year, we come 

 nearer to the conditions observed with respect to the other 

 epochs than we should do by taking 1S48. 



Before proceeding to deduce a few results from Mr. Rawson's 

 valuable " Report upon the Rainfall of Barbados " from 1843 to I 



1871, with a copy of which he has favoured me, I would remark 

 that he has made apparently some oversights in his letter. For 

 example, he says, with reference to a comparison of the rainfalls 

 at Fairfield and Ilalton, "but the rainfall at Fairfield during 

 the last three years . . . is 13-33 per cent, below that of Haltor. 

 Therefore 217 in. have to be added to the minimum average of 

 1S43-45, which would increase the above excess of io-6 in." 

 But if the minimum average be increased by a percentage, 

 would it not be well to increase also the maximum average of 

 1 84 7-49 by the same percentage? If this be done, the excess is 

 not altered in the least. 



The earliest rainfall observations at Barbados, given by Mr. 

 Rawson, were those taken at Fairfield from 1843 to 1850, after 

 which there is a long blank. Now, the rainfall there during 

 that period gives the following results : — 



M 



1843-45 1637 



Ma 



years. Rain. 



1S4S-50 179.7 



showing an excess of 16 inches in the maximum period. 



The next earliest and most complete observations are those 

 taken at Husbands ; they commence with 1847, and have been 

 continued without interruption. From them we get : — 

 Min. yc 



184S-50 182-3 



1859-61 183-3 



36v6 



1S55-57 1S8-1 



1866-68 162-8 



35o'9 



which gives an excess of 14-7 inches on the maximum side. 



The greatest number of inter-comparable observations for the 

 longest period are those taken at the eight stations, Binfield, 

 Ilenly, Husbands, Grand View, Oughtersons, Ilalton, Fdge- 

 cumbc, and St. Ann's, from 1855 to 186S ; and I find that they 

 give a mean excess of 56-9 inches on the side of the years of 

 maximum sun-spot. 



I do not think that these results are opposed to the hypothesis 

 which Mr. Lockyer and myself have put forward. As a matter 

 of fact, the rainfall of Barbados, as given by Mr. Rawson from 

 184310 1868, bears out the hypothesis if we take 1849 as a 

 middle maximum in place of 1848 ; and it is for others to judge 

 whether the reasons that have been assigned for the change from 

 184S to 1849 (not for Barbados alone, but generally) are valid. 



But it maybe said that the rainfall of 1871-73 was opposed 

 to the hypothesis. I have not the rainfall for those years before 

 me. Granting, however, that they show a very considerable 

 diminution, the question arises whether the favourable result of 

 twenty -six years (1843-6S) are to be upset by the unfavourable 

 results of three years (1S71-73) ? Have we not in meteorology 

 many such exceptions to well-established laws ? 



The rainfall at 250 stations in different parts of the world has 

 now been examined, and the results are so decidedly favourable 

 that it is practically of no consequence whether the experience of 

 Barbados is for or against the theory. I think the more the sub- 

 ject is examined, the more clearly will the law come out ; but we 

 must be guided by facts, and not hesitate to discard this or any 

 other theory when unsupported by facts. 



Mauritius, Oct. 15 C. Meldrum 



Ice-Caves 



The occurrence of snow and ice in an old mine during the 

 month of June, mentioned by Mr. J. Clifton Ward in his in- 

 teresting paper in N.\ture, vol. xi. p. 309 — to the accuracy of 

 the greater part of which I can bear personal testimony — has a 

 more exact parallel in the Alps than "a Swiss glacier," n.amely, 

 a glaciere. These remarkable caverns have been fully described 

 by Mr. G. F. Browne in his able and pleasant work, " Ice- 

 Caves of Switzerland and France;" and briefly by myself 

 in "The Alpine Regions." Since the pubhcation of th.it book 

 I have seen others ; and as one of these has never, I think, been 

 described in any English work, I venture to t.ake the oppor- 

 tunity of sending you a short account of it. It is in the Val 

 d'Herens, a short distance from Evolena, on the way to the Pic 

 d'Arzinol, and is called the Pertuis Freiss. A slip or subsidence 

 of part of a cliff appears to have cracked the rock and opened 

 two joints, into one of which fissures one can descend. This is 

 about four feet wide and generally some four yards high, the floor 

 being a little below the level of the ground outside. Tlie crevice 

 comes to an end in about a dozen yards. Against the slightly 

 sloping wall of rock rested some pendent sheets of ice, whose 

 thickness rarely appeared to exceed three inches, anil irregular 

 patches of ice lay about the floor. The temperature of the air 

 appeared to be a little above the freezing (unfortunately, I had 



