Of course it was not to be expected that the results of a single year should agree exactly with the mean of two other years, still less when the size of gauge used was different, and the locality so opposite as the inland district of Calne and the rock-bound Yorkshire coast. We therefore look upon it as satisfactory that in only four months out of eleven do the ratios at Calne and Hawsker differ more than 3 per cent. In April, June, and November they are identical. The Calne results are thus strongly confirmed; and it may be considered as certain that pit-gauges always exceed those at one foot, although the precise amount of excess remains to be determined. In our last Report we expressed the hope that we should this year be able to state the result of the discussion of all the rainfall registers which were absolutely continuous from January 1, 1860, to December 31, 1869. We have the pleasure of doing so in two respects, viz. (1) with reference to their bearing on the question of the existence or otherwise of secular variation of rainfall in the British Isles, and (2) as data indicative of the distribution of rain over the country. The secular variation of rainfall, or the relative dryness and wetness of different years and groups of years, is one of the most important and difficult branches of rainfall work. It has been treated in our Reports for 1865, and very fully in that for 1866. In the latter we gave the calculations in detail, from which the values shown on the accompanying diagram were obtained. Referring to that Report for full explanation, we have only now to mention that the subsequent years 1866 to 1869 have been computed in the same manner and added to the diagram (fig. 1). We may also remark that various observations collected since its publication have confirmed the general accuracy of the curve quite as much as could have been anticipated. On the present occasion we do not intend to discuss the relative rainfall of different years, but the relation of the fall during the ten years 1860-69 to previous decades. For this purpose we have grouped the yearly values in decennial periods, similar to those adopted in our 1867 Report, whence we obtain the following result: TABLE II.-Ratio of Rainfall in each ten years since 1730 to the Mean of sixty Years, 1810-69. Having previously pointed out the peculiarities of the earlier portion of the curve, it is only necessary on the present occasion to call attention to the last forty years, whence it will be seen that, according to this mode of investigation (which is principally based on English returns), three out of the four decades had a rainfall nearly identical, and the other (1850-59) considerably below them, the deficiency being nearly 7 per cent. This result is based on a combination of records, as fully explained in our 1866 Report. We proceed to examine how far it is corroborated by individual stations, but are at once confronted by the paucity of stations of which perfectly continuous records for even half a century exist. We therefore confine ourselves to the forty years, from 1830 to 1869, for which period we have twelve perfect records at widely separated stations. The mean fall in each decade and in the whole period, and the ratio of each decade to the whole period at each station, is given in Table III. From careful examination of Table III., it appears that the amount of rain which fell in the ten years 1830-39 was very similar to that which fell in the ten following years, the difference being a decrease, but scarcely one per cent. The investigation in our 1866 Report shows an increase of 1.2 per cent.; and examination of returns ceasing in 1850, and therefore not quoted in either Report, show several cases of absolute identity. With one investigation leading to a decrease of 1 per cent., another to an increase of the same amount, and a third to identity, we are led to the conclusion that the two decades may be considered to show similar results. This is a much more important fact than it at first appears; and for this TABLE III.-Comparison of the Rainfall in each Decade since 1829 with the Mean Rainfall of forty years, ending with 1869. reason while there are only about a dozen registers complete for the four decades, there are thirty-eight which are complete for the last three decades. Now that we have found the relation between the first two decades, the returns for the thirty years are rendered almost as instructive as those for forty years. We have therefore compiled Table IV., which differs from Table III. only in its being for thirty years instead of forty, and in giving observations from thirty-eight stations instead of twelve. |