418 



SCItlNCE. 



[N. S. Vol XV. No. 376. 



sures observed on the Rocky Mountain 

 plateau to sea level is a problem of great 

 scientific difSculty for two reasons: (1) 

 Becaiise it is not evident what the exact 

 effect of the plateau is as modifying the 

 ordinary Laplaeean free air reduction, 

 even when the mean temperature of the 

 air column o is assumed; and (2) because 

 the vertical temperature gradient between 

 the observed surface temperature t and 

 the sea-level temperature <„ is difficult to 

 determine, or in other words because the 

 connection between \ (t-\- '!■!,) and the true 

 ti is hard to find. The solution of this 

 problem has been forced upon the Wash- 

 ington Office ever since the opening of the 

 service in 1870, inasmuch as the results of 

 such reductions are used to form the daily 

 weather maps, and the errors of reduction 

 result in inaccurate systems of isobars, and 

 consequently incorrect deductions regard- 

 ing the existing weather conditions, es- 

 pecially west of the Mississippi Valley. 

 The perplexity of the problem may be in- 

 ferred from the fact that the present sys- 

 tem constitutes the sixth effort to solve it. 

 The indications are that the new isobars 

 conform very closely to the true weather, 

 and that the practical working of the sys- 

 tem will prove to be satisfactory to the 

 Bureau. 



A brief summary of the earlier methods 

 of i-eduction is as follows: 



1. For the years 1870-81 all the low sta- 

 tions wei'e reduced by an elementary ap- 

 plication of tables given in Loomis ' ' Me- 

 teorology' (the same as Giiyot's table D, 

 XVI., edition of 1859), using the temper- 

 atures observed at the moment of observa- 

 tion. 1872-1880. The higher stations 

 were not reduced except by the applica- 

 tion of a constant appropriate to the mean 

 annual temperature and pressure ; the low- 

 er stations continued with the elementary 

 method, but this included an erroneous 

 use of the observed pressure. 



2. 1881-85. The Abbe-Upton system of 

 monthly constants for each station, based 

 on the mean monthly temperatures and 

 pressures. 



3. 1886-1887. Ferrel's system was in 

 operation but found to be in a form which 

 was too complicated for practical station 

 work. However, some correct principles 

 were introduced by him, namely, the mean 

 temperature of twenty-four hours as the 

 argument instead of the temperature at 

 the time of observation ; a separate correc- 

 tion for the plateau effect; the variation 

 of the reduction Avith the local pressure j 

 and a correction of the surface tempera- 

 ture t to the mean by an approximate 

 'vertical gradient. 



4. 1888-1890. A mixed system, partly 

 Ferrel's and partly Hazen's. 1891-1901. 

 Hazen's empirical system alone, in which 

 numerous changes were made in the wrong 

 direction. The plateau correction was 

 omitted, the pressure argument was aban- 

 doned, and the surface temperature not 

 corrected to e became the only argument. 

 In constructing the empirical tables it was 

 assumed that Mt. Washington is a correct 

 type for the plateau effect, which is not 

 true; the pressure on the sea level was 

 taken as exactly 30.00 inches in working 

 out the reductions, which does not conform 

 to the facts; the check upon the reduction 

 was limited to the criterion that the iso- 

 bars could be smoothly drawn, and reduc- 

 tions for many stations received arbitrary 

 modifications for that purpose. The result 

 of this system was to give too high sea- 

 level pressures at low temperatures, espe- 

 cially in cold waves, and too low pres- 

 sures in warm weather. 



5. In 1896 Professor Morrill computed 

 tables which have been used somewhat in 

 the office, but never published, in which 

 the mean temperature 8 and the pressure 

 arguments were restored, and the treat- 

 ment of the humidity put on an improved 



