Statistics also of the annual amount of rainfall became of commercial value as bearing on questions of the economic supply of water for large towns, and much valuable information was acquired as to the dependence of mortality on different kinds of weather. Of more purely scientific interest were the variations of pressure, temperature, wind, etc., depending on the time of day, or what are technically known as diurnal variations, which were brought to light by these comparisons. This branch of the subject is known as “Statistical Meteorology,” and has advanced very little since it was first developed by Dove and Kaemtz. When the attempt was made to apply statistics to weather changes from day to day, it was found that average results were useless. The mean temperature for any particular day of the year might be 50°, if deduced from the returns of a great many years, but in any particular year it might be as low as 40°, or as high as 60°. The first application of the method was made by the great Napoleon, who requested Laplace to calculate when the cold set in severely over Russia. The latter found that on an average it did not set in hard till January. The emperor made his plans accordingly; a sharp spell of cold came in December, and the army was lost. It has now been thoroughly recognized that statistics give a numerical representation of climate, but little or none of weather, and that large masses of figures have been accumulated, to which it is difficult to attach any physical significance. The misuse of statistics has done much to bring the science of meteorology into disrepute.