REPOET OF THE SECRETARY. 79 



wave length 0.375yu to wave length 2.5/^ in about twenty-five minutes of time. This 

 region extends from beyond the line "L," or farther than the eye can see without 

 siaecial means in the ultra violet, through the whole visible spectrum, and on through 

 the visible but very intense upper infra-red spectrum as far as glass is transparent. 

 It includes about j^^'^% of the solar radiation which reaches the earth's surface, and 

 so far as experiment has shown, within 1 and 2 per cent of all that reaches the outer 

 layers of the earth's atmosphere lies within this spectral region. 



It is only since January, 1903, that the apparatus has been so far perfected as to 

 include in the regular ])olographic work the important portion lying between 0.375// 

 and 0A7u, and in Plate V readers may see for the first time, as the bolometer recog- 

 nizes them, the general features of the violet solar spectrum so familiar in photo- 

 graphic spectra. It is, of course, impossible to show the finer details when the bolom- 

 eter passes through the whole visible and upper infra-red spectrum in less time 

 than was occupied in passing from through a fourth of the upper infra-red alone in 

 preparing the detailed map published in 1900; but nevertheless in a rough compar- 

 ison of three curves it was seen that as many as 325 of the Fraunhofer lines were 

 discriminated by the bolometer as it passed over them thus rapidly. As remarked 

 last year, scarcely any "drift" of the galvanometer is now experienced, and, indeed, 

 it is sometimes possible to take holographs for a month without readjusting the 

 bolometric circuit in any way. This excellent behavior is principally due to the 

 improved rheostat and to the 16-coil type of galvanometer, both of which were men- 

 tioned on page 87 of last year's rejjort. 



Transmission of the atmosphere. — From series of such holographs as are described at 

 page 89 in my last year's report, coefficients of transmission of the atmosphere are 

 obtained. It is now customary to compute them for more than 30 points in the 

 spectrum between wave lengths 0.37 m and 2.5 //, of which 24 are at wave lengths 

 where there are no prominent atmospheric bands and the others within such bands. 

 The reduced observations take such a form that they may be graphically platted as 

 straight lines whose angle of inclination is a measure of the transmission coetficient 

 of the air at the given wave length. It is the exactness with which the reduced 

 observations from the holograph fall upon such straight lines which furnishes the 

 principal criterion of their value. To show how close this agreement is on the best 

 days, I invite attention to Plate VI, which gives for several wave lengths the reduc- 

 tions of the observations of March 25 and 26, 1903, respectively. Ordinates are 

 logarithms of heights of the holographic curves at the selected wave lengths and 

 abscissfe are air masses. 



The circles represent observations of March 25 and crosses those of March 26. 

 Lines I and II are for a wave length of 1.027 //; III and IV at 0.656 /<; V and VI at 

 0.468 jii, and VII and VIII at 0.395 u. On nearly all days of observation it is found 

 that the forenoons yield a less regular series than the afternoons, and it appears as if 

 the air became clearer and clearer till a little after noon and then remained substan • 

 tially unchanged on the best days till 4 or 5 o'clock p. m. It is therefore the practice 

 here to use only afternoon observations in determining atmospheric transmission. 

 The forenoon observations are distinguished in Plate VI by being connected by dotted 

 lines. 



Notable decrease in tJie transparency of the air in the present calendar year. — From 

 numerous determinations of the transmission of the air for solar radiations, a 

 striking diminution of the transmission has been noted in the best days of this year 

 as compared with last year. To illustrate this difference 1 give the following table: 



