292 F. II. Bigeloiv — The Earth's Atmosphere. 



aqueous vapor in the great currents of the circulation. It has 

 been customary to take the mean of a series of years or months 

 for the normal, and the difference between this normal and the 

 individual values for the departures. Fig. 4 shows the general 

 case that is to be considered, the data being taken from Abstract 

 3 of the Climatological Division. The normal of precipitation 

 for the Lake Region and Central valleys, 1872-1907, is 35*55 ; 

 for .the east Gulf States it is 50*04; for the south Pacific 

 States and southern Rocky Mountain Plateau it is 14*55 inches 

 per year. If the normal line on fig. 4 lies between the two 

 close parallel lines, the departures are measured from this nor- 

 mal to the crests of the annual line ; the residuals would be 

 measured from the consecutive mean line to the crests. When 

 the normal and the consecutive mean line are practically coin- 

 cident as in the section 3, the departures and the residuals 

 agree together, and no question arises. When, as in sections 1 

 and 2, the normal line and the consecutive mean line do not 

 agree, then the residuals are different from the departures. In 

 such a case as the east Gulf States for 1872-1889 the entire 

 system is above the normal, and for 1890-1907 it is entirely 

 below the normal. It does not seem proper to confuse residuals 

 with departures for practical purposes, because a series of de- 

 partures all largely positive for a set of years, and all largely 

 negative for a following set of years, does not give a good idea 

 of the course of the phenomena in the current years that are 

 within the memory of man. 



It is not easy to see how the line of consecutive five-year 

 means can be extended forward with such accuracy as to make 

 it suitable for a computation of the residuals of the current 

 month and year in which the Climatological data are to be 

 published. It is proper to submit this question to public dis- 

 cussion, and an expression of opinion is solicited from engineers 

 regarding the desirability of substituting approximate residuals 

 for the wide departures at present in use. The question is not 

 concerning the causes of the change in the normal precipitation 

 in different series of years, such as change in the general cir- 

 culation, change in the exposure of the rain gauges due to the 

 growth of cities, nor of the effect of deforestation and cultiva- 

 tion of the soil upon climate. It only concerns the treatment 

 of the records of precipitation as they are actually made. It 

 is obvious that the departures of precipitation differ materially 

 in principle from the departures of temperature as published 

 in the Monthly Weather Review, and its readers should at 

 least be aware of the problems concerning their discussion. If 

 the printed departures stand as raw material without fixed 

 principles of interpretation, it is evident that differences of 

 opinion will arise regarding practical results which are due 

 solely to different ways of interpreting the facts of observation. 



