C. s. Cook— Spectrum of Aqueous Vo/por. 268 



all, the vapor contained in the stratum of 3,000 feet which we 



arc studying becomes a Bmall factor in the whole. The ob- 



\ w at the top of the mountain we Buppoee has several miles 



ot' damp air to look ap through. The observer al the base has 



this same distance i>( several miles plus 3,000 feet. The 

 Bpectra will not be proportionally very different. We repeat 

 then that the minimum difference between the spectroscopic 



observations at the top ami bottom of the mountain will be 

 observed when the vapor extends to the greatest height in the 

 atmo>phere. This will naturally he during the progress of a 

 storm. It is found that such is the case. 



In looking over the records, I find the difference very gen- 

 erally small in threatening weather, becoming sensibly in 

 of actual storm with precipitation. 



In only one case is the reading at the upper station greater 

 than a simultaneous one lower, S units in this case. This 

 represents an impossibility, and means simply an error in the 

 reading on the part of one of the observers. A difference of 

 1 to 3 units is not considered significant, in view of the very 

 considerable difficulty of such work. When we consider that 

 the two observations were, from the circumstances of the case, 

 made absolutely independently, with constant and large fluctu- 

 ations in the readings, and that the observation is one of great 

 delicacv, we must consider this result a very gratifying one. 



Let us see what light is gained from averaging the observa- 

 tions as far as possible. (We are obliged to exclude many 

 observations mainly on account of the interference of clouds.) 

 We have seen that the difference is a maximum when the line 

 is at the top, i. e., when the absolute humidity is least. I 

 assume a vapor tension of 04 inch as the average humidity at 

 that altitude, and find the mean of all spectroscopic differences 

 when the vapor tension was less than this ; and also in the same 

 way when the tension is greater than this. With a tension of 

 more than *4, the mean difference (17 readings) is 8*5 units. 

 With a tension of less than '4, the mean (1-4 readings) is 15*7 

 units — a ratio approaching tw^o to one, and amply sufficient to 

 establish the point. The same thing is illustrated by a diagram. 

 Fig. 2. It compares the same two quantities, having a curve 

 for spectral difference, * and one for the summit humidity. The 

 two should correspond. A third curve is added which should 

 be antithetical. This curve represents the difference of vapor 

 tension at the base and summit ; in other words the humidity 

 gradient. If the humidity falls off but little in ascending the 

 mountain it seems probable that one would have to go to a 

 great height in the atmosphere to reach a point of low 



* This is the lowest of the curves on fig. 2. 



Am. Jolu. Scl— Third Series, Vol. XXXIX, No. 232.— April, 1890. 

 18 



