REPOET OF THE SECRETARY. 



89 



where (A) is an inytrumentaliactor connected with the vvidtli of the sHt, the absorji- 

 tion of the optical surfaces, the (luality of the bolometer, and the sensitiveness of the 

 galvanometer. We may assume {rj and (k) to be constant for the short si)ace of 

 time occupied by one day's observations. We then obtain the following equation: 



log. {d)=myr log. (rt)+ log. (Av',j) 



of which the last term is I'onstant. This etjuation is in the form of the ecjuation of a 

 straight line, so that if we make a plot in whicii heights are proportional to 



T> 



logarithms of ((/), and horizontal distances to values of (m=f> ), the various observa- 

 tions should determine a straight line the tangent of whose inclination is the quan- 

 tity log. («). 



The procedure has l)een followed with many of the energy curves already taken, 

 and from results of four of the clearest days I select the following data for a few 

 wave lengths, graphically rei)resented in Fig. 'A. The coetlicients of transmission are 

 the jiercentages of the solar radiation of the given wave lengths which are transmitted 

 by a vertical colunni of air at standard ])arometi'ic i)ressnre. It is not the selective 

 transmission in bands, but the general transmission at the given wave lengths which 

 is here in (juestion. 



Fig. '■'>. — The transparency of the earth'.i atmospliere. Horizdntiil scale represents wave lengths. Ver- 

 tical scale exhibits proportion of incident light transmitted ))y vertical colninn of air at standard 

 barometric pressure. 



