420 



SCIENCE. 



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



to 0.400 inch, should be reduced to that 

 which is observed at the low-level stations, 

 about 0.150, by a correction of the form 

 C. Ae.IL, where C=0.00100, M is the 

 departure of the temperature from the 

 annual mean, and E is the elevation in 

 units of a thousand feet. This is readily 

 computed for each station, and it is to be 

 added to the free air correction computed 

 by the Laplacean formula with modern 

 constants. 



The monthly station pressures B were 

 now reduced to sea level, giving B^, and 

 isobars were drawn as well as possible 

 through the resulting values. Many of 

 the old stations had their elevations deter- 

 mined only by barometers and were quite 

 erroneous; in many cases the temperature 

 argument used J (t-\-ia) was not exact 

 •enough to give very accurate results; not 

 ■a few stations had only a short series of 

 .years to use in constructing their normals ; 

 from these causes considerable irregularities 

 were found on the first system of sea-level 

 maps. The pressures for each station were 

 now interpolated from the map, B„, and 

 ,the differences, B^-B^^^, computed. For 

 ^certain stations these differences were 

 about constant, indicating an error in the 

 ;adox)ted elevation, or in the mean tfemper- 

 ature from which the plateau effect was 

 reckoned; in other cases the differences 

 had a variation in an annual period, show- 

 ing that the true value of 6 differs frord 

 1 (t-\-to)- By readjusting our data to 

 allow for all these considerations, the sea- 

 level pressures were computed a second 

 time. The differences, B^^-B^, were now 

 quite small for stations of long record, us- 

 Tially less than 0.010 inch. Assuming that 

 the normal values of the short record sta- 

 tions should be reduced to the long record 

 : series, that is, 20 to 27 years, these last 

 residuals were added to the original sta- 

 tion pressures B to give the station homo- 

 geneous normals B . There still remain a 



few stations, some of them at low level, so 

 that any adopted method of reduction can- 

 not be a possible source of error, wherein 

 a nearly constant residual reduction is yet 

 required to reduce them to the homogene- 

 ous system, marked J A. This is probably 

 due to some local peculiarity of the wind 

 circulation, or the exposure of the barom- 

 eter, and it may properly be considered as 

 a topic for further investigation. 



We next proceeded to make reductions 

 for all the stations now in operation to 

 the 3,500-foot plane and the 10,000-foot 

 plane, both of which are useful in the 

 studies of cyclones and anticyclones, but 

 instead of directly from B to B^ and B^, 

 by a roundabout circuit. The sea-level 

 values B^^ . t^. e^. pressure, temperature, 

 vapor tension, were interpolated on the 

 centers of reduction from the sea-level 

 charts. The temperatures t-^ . to and the 

 vapor tensions e^ . e„, on the two upper 

 planes, respectively, were computed by gra- 

 dients derived from the cloud computations 

 of 1896-97, and balloon and kite ascensions. 

 With this data charts of t^.t.^.e-^.e^. were 

 formed, the pressures B^.B^. on these 

 planes were computed by means of our 

 new logarithmic general tables, and the 

 corresponding pressure charts drawn. 

 Thence the station data B^ .t-y.e^. for the 

 3,500-foot plane and B^.t^.e.,. for the 

 10,000-foot plane were interj^olated for 

 each month and for the annual mean. As 

 a check, station data S„ .t.e. were reduced 

 so as to give corresponding to £„ the values 

 of Bj and B^- We have thus derived 

 i?i . Bo. by two separate methods, and they 

 generally agree to about 0.01 inch on the 

 average. This check includes the construc- 

 tion of the general tables and the special 

 station tables, also the drawing of the dif- 

 ferent sets of charts. We have therefore 

 obtained the same results by means of two 

 paths of reduction-, the first to sea level 

 from the station, and thence through the 



