METEOROLOGY. 801 



placed upon an axis and at a uniform distance of 20 3'ards from the ob- 

 server. {Z. 0. G. M., xviii, p. L'ST.) 



154. G. M. Wliiijple, of Kew, i^roposes to apply Sir Francis Galton's 

 method of comjiosite portraiture to the stud}' of meteorological observa- 

 tions. Thus, by selecting a large number of typical storm-charts and 

 sui^erposing their prominent features upon one chart, we may finally 

 obtain a tracing that may be considered as an average normal presenta- 

 tion of facts. [This idea of saving the labor of numerical compilations 

 has been independently applied to the same study by W. M. Davis, of 

 Cambridge, Mass., who has thus thrown upon one chart the results of 

 a large number of observations bearing upon tornadoes.] 



155. Prof. K. Weihrauch, of Dorpat, has studied the proper method of 

 taking the arithmetical mean of a series -of values of the relative hu- 

 midity. The arithmetical mean is always proper when the quantity is 

 directly measured and has no algebraic connection with other objects of 

 observation ; this is true of temperature, absolute humidity, ])ressure, 

 «S:c. But the relative humidity is the ratio of two numbers, and the 

 average value of these ratios is not the same as the ratio of the average 

 values. The numerators and denominators of the fractions that consti- 

 tute relative humidities have each regular diurnal periods, and the diur- 

 nal i^eriodicity of the relative humidity is therefore a complicated func- 

 tion of the other two. The true average humidity for the day can only 

 be found by knowing the true average temperature of the air and abso- 

 lute humidity, and it will differ therefore from the simple average of a 

 few observed relative humidities. In general Weihrauch finds that the 

 true relative humidity is less than the arithmeticar average, and the 

 true amplitude greater than the arithmetical difference. Ue shows that 

 it is necessary for us to give up taldng the arithmetical mean relative 

 humidity on account of systematic errors, which all lie in one direction. 

 We must undertake a slightly greater labor, namely, the addition of the 

 maximum vapor tension to each table of daily temperatures, with which 

 data the true mean humidity can easily be calculated by the method 

 which he gives. The hourly, daily, monthly, and yearly mean maxi- 

 nmm vapor tension should also be computed for sake of the convenience 

 with which we then compute the true average for long periods. {Z. 0. 

 G. M. XIX, p. 2G5.) 



156. [Promptedby the many troubles which beset the use of relative 

 humidities, and on account of the decided direct advantage of using 

 the dew-point in the prediction of frosts, rain, snow, fog, &c., the Sig- 

 nal Office has for some years introduced this latter datum into its sta- 

 tistical tables and daily weather maps. As the dew-point is a matter 

 of direct observation, and has no necessary relation to temperature, its 

 mean value can be taken Avithout incurring the objections urged by 

 Weihrauch against the relative humidity. If, however, we desire to 

 compute the mean absolute or relative vapor tension from the monthly 



