L. Bell — Rainband Spectroscopy. 351 



much blacker than was ever necessary. Consequently a small 

 rotation of the Nicol would duplicate the strongest rainband 

 and the scale was somewhat limited in extent. The tube car- 

 rying the Nicol was graduated to give a scale of about eighteen 

 grades of intensity, but for the above reason only eight could 

 be used, which however proved sufficient to give very good 

 results. The rainfall and relative humidity for the period cov- 

 ered by the chart were obtained from the Signal Service sta- 

 tion at Baltimore. In the chart the upper curve gives the 

 variations in y from day to day, the arrows being placed to 

 indicate the time when a rain began, and the amount of rain- 

 fall being annexed. The dotted curve below shows the per 

 cents of relative humidity for the same period. During the 

 time represented there were nineteen rains, of which number 

 fourteen (73 per cent) were distinctly predicted from two to 

 thirty-six hours before their occurrence by an increase in the 

 intensity of the rainband. On grouping the rainfalls in the 

 case of a continued storm, like that of the last week of March, 

 there were fourteen rains, of which thirteen (93 per cent) were 

 predicted. In only one case did a rain come entirely without 

 warning — the shower of March 18, which drifted up suddenly 

 between observations. In one or two other cases rain fell 

 shortly after the morning observation ; but on a careful review 

 of the observations, it appears that during the entire six months 

 about eighty per cent of the rains gave fair warning through 

 the spectroscope. This is a very satisfactory degree of effi- 

 ciency for a small instrument. It will be noticed that in the 

 chart there appears no very obvious connection between the 

 strength of the rainband and the amount of rain. Now and 

 then there is a rough correspondence, but that is all. A good 

 reason for this is found in the fact that now and then an obser- 

 vation would be taken during a previously predicted storm 

 and thus y would be abnormally strong, as it is generally dur- 

 ing a rain. The curve of relative humidity corresponds quite 

 closely with the rainband curve, though in general its maxima 

 and minima are some hours later. The occasional discrepan- 

 cies are undoubtedly due to the fact that while the hygrometer 

 indicates only the lowest layer of aqueous vapor, the spectro- 

 scope sums up everything along the line of sight and thus 

 indicates vapor in the upper strata of the atmosphere entirely 

 out of reach of the hygrometer, while on the contrary an earth 

 stratum quite dense enough to affect the latter instrument 

 would be frequently too slight to make itself apparent as a 

 rainband, particularly in a small instrument. 



In this integrating power of the spectroscope lies its strength, 

 and unfortunately also its weakness as a meteorological instru- 

 ment. While it shows with considerable accuracy the total 



