S. P. Langley on the Selective Absorption of Solar Energy. 159 



did, were found to indicate a probble error, for a single obser- 

 vation, of less than 1 per cent. 



The variations from minute to minute (under a visually clear 

 sky) amount frequently to ten times the probable instrumental 

 error; and they can only be eliminated by repeating the obser- 

 vations a great number of times on many different days. 

 Actually twenty-nine such days' observations have been made 

 (as appears below) in the preliminary series ; but it would be 

 an error to suppose that this number was obtained without the 

 sacrifice of a still larger number on which the apparatus was 

 prepared and the day spent without results, owing to the still 

 more considerable atmospheric changes between morning and 

 afternoon. Even of the twenty-nine days cited, and which 

 may be considered exceptionally fair, it will be seen that in 

 only ten cases did the sky continue sufficiently constant in the 

 morning and afternoon to allow complete series to be taken. 



It will be understood that we aim to make at least two sets 



of measures throughout the spectrum daily — one when the rays 



have been little absorbed (at noon), the other when they have 



been greatly absorbed (in the morning or afternoon). The 



mass of air through which the rays pass is taken proportional 



to secant f for zenith-distances less than 65°, and for those 



'01 74 tabular refraction , . , ., ., ., 



greater to ; — r-r-, — -, — , and m both cases to tne 



° cos apparent altitude 



actual barometric pressure. It is expressed in units, each of 

 which is represented by the pressure of one decimetre of mer- 

 cury at the sea-level. As the barometric pressure there is 

 7'6 decim., t 7 ' d gives the transmission for an entire atmosphere. 

 The coefficient of transmission, then, for one atmosphere (^' 6 ) 

 is the proportion of the radiation transmitted by a sun in the 

 zenith to an observer at the sea-level; and this is here show-n 

 to vary greatly for each ray. Thus, by reference to Table III., 

 we find of three solar rays whose wave-lengths are "375, '600, 

 l - 000, that of the ray whose wave-length is 0^*375 (in the 

 ultra-violet), 61 per cent, of the original energy w^ould be ab- 

 sorbed and 39 transmitted; of wave-length 0^600 (in the 

 orange), 36 per cent, would be absorbed and 64 transmitted ; 

 of wave-length 1^*000 (in the infra-red), 20 per cent, is ab- 

 sorbed and 80 transmitted, &c. 



Allegheny Bolometer Observations on the Solar Diffraction- 

 Spectrum previous to Mt. Whitney Expedition. 



The following list shows the dates at which bolometer ob- 

 servations were made at Allegheny up to June 1881, for the 

 measurement of heat in the spectrum and the determination of 



