March 14, 1902.J 



SCIENCE. 



419 



basis, though the plateau effect was 

 omitted and the correction between t and 

 e only roughly determined. 



The following statement will indicate 

 the most important changes which have 

 been recently adopted. The principles in- 

 troduced by Ferrel have been more closely 

 followed than any of the others, but de- 

 cided improvements have become possible 

 by reason of the gradual accumulation of 

 accurate observations on the plateau. It 

 was first necessary to construct exact nor- 

 mals of station pressure. There have been 

 numei'ous changes in the location of the' 

 offices during the past thirty years, involv- 

 ing variations in the elevation; there has 

 been a gradual improvement in the gen- 

 eral national surveys for elevation by 

 which the local bases can be referred to 

 the sea level; the instrumental errors were 

 neglected during certain years when less 

 than, ± 0.007 inch; the observations have 

 been made at different sets of selected 

 hours ; and the gravitation correction was 

 not regularly applied. To reduce the en- 

 tire set of observations from 1873 to 1900, 

 inclusive, to a homogeneous system, they 

 have been corrected to the elevation adopt- 

 ed for January 1, 1900, or the one occupied 

 nearest that date, also to the mean of 

 twenty-four hourly observations, and the 

 corrections for instrumental errors and 

 gravity have been added systematically. 

 The monthly and annual means give the 

 normals, and from these the variations in 

 the year and from year to year are com- 

 puted, the latter becoming the basis for the 

 further discussion of climatic and seasonal 

 problems. 



The process of determining the sea-level 

 temperature beneath the plateau was con- 

 ducted as follows: The normal mean 

 monthly temperatures of all stations be- 

 tween the Pacific coast and the Mississippi 

 river were collected by groups accordiag 

 to their elevations, and reduced to selected 



planes through short distances. Thus all 

 temperatures observed between and 

 1,000 feet were corrected to the 500-foot 

 plane, between 1,000 and 2,000 to the 

 1,500-foot plane and so on up to 7,000 feet. 

 Temperature gradients in latitude and 

 longitude were worked out by discussing 

 these data, and then all the data on the se- 

 lected planes were further corrected to 

 values on the centers of reduction, that is 

 the iDoints where the five-degree meridians 

 and the five-degree parallels intersect. The 

 several stations were carried in various di- 

 rections to different centers, so that purely 

 local conditions might neutralize them- 

 selves. Over these centers we have thus 

 formed from the observations different 

 temperatures in a vertical direction, and 

 they were then plotted as points on dia- 

 grams through the average of which ver- 

 tical temperature curves were drawn, and 

 prolonged to sea level without much error. 

 These sea-level temperatures were now 

 transferred to charts of the United States 

 and Canada, and in connection with all 

 the stations available from the Atlantic to 

 the Pacific coasts, sea-level isotherms were 

 readily drawn, by which all the minor dis- 

 crepancies occurring in the plateau dis- 

 trict were removed. By interpolation vve 

 then found the true terminals of the ver- 

 tical temperatures at sea level on the cen- 

 ters of reduction. This entire work was 

 performed two or three times in succession 

 by a series of approximations, and the in- 

 terlocking of the vertical and horizontal 

 lines on the sea-level plane M^ere made to 

 conform to the observed conditions. 

 Thence the sea-level temperatures for the 

 respective stations were found by inter- 

 polation from the isotherms to tenths of a 

 degree, so that we have thus accurately ob- 

 tained tg as well as t. 



The plateau effect was determined on 

 the theory that the wide s\^fing of the tem- 

 peratures in the annual period, amounting 



