WHOLE VOL. NOTES. SOUTH AMERICA 



59 



and the coast station of Iguape, the last in a strategic position, 

 owing to the circumstance that deviations from the normal 

 in the volume, intensity, and direction of the Brazil current 

 render the data of much importance in world weather prob- 

 lems (see Quart. Journ. Roy. Meteorol. Soc. London, vol. 

 45- P- 59> Jan. 1919). 



sao paulo, capital 

 Temperature. 



Observations from 1848 to 1858 were made by Brigadier Jose 

 Machado de Oliveira in ° F. at the hours of 6 a.m. and 

 3 p.m. The means appear in a table, page 9 of the " Dados 

 Climatologicos dos Annos de 1887 e 1888" (Sao Paulo, 

 1889). They were first converted to ° C. and then brought 

 to mean of day from a table given in the " Dados Climato- 

 logicos do Anno de 1901 " (Sao Paulo, 1902), page 120, 

 based on hourly means referring to the period 1889 to 1901, 

 the following being the corrections applied : 



°c. 



A critical examination of these data in comparison with the later 

 Sao Paulo series showed that they were about 1 ° C. too high 

 on the mean of the year, so that Machado de Oliveiras' 

 monthly means were corrected by assuming a minus correc- 

 tion of i°. The corrected values may be taken as close 

 approximations. 



Observations began in Sao Paulo in 1886, being first made in a 

 private residence, height 730 m, and then in March 1888 in a 

 station established in the Jardem da Luz, height 740 m 

 until 1894, when the station was transferred to the Normal 

 School, height 755 m, later 761 m, and in August 1902 a 

 new station, height 815 m, was put into operation in the 

 Avenida Paulista, the new Observatory built in 191 2, height 

 820 m, being contiguous, so that since 1903, what may be 

 considered a homogeneous series of data are available. 



The data up to 1902 were derived from the observations taken at 

 the Normal School, which continued to function until about 

 1920. These data, used up to 1902, were corrected as follows, 

 from a comparison based upon synchronous data from 1903 

 to 191 2, which give the following corrections to reduce the 

 earlier data to 815 m: 



