194 BOTANICAL GAZETTE [SEPTEMBER 
capacity of 500 cc., closed with tightly fitting cork stoppers that 
were perforated for the atmometer tubes and for bent capillary 
glass tubes which served to equalize the atmospheric pressure within 
the bottles with that of the exterior air, without causing any loss by 
evaporation or permitting rain water to enter the reservoir. The 
bottles were sunk in the soil about two-thirds of their height, so 
that the evaporating surface of the instruments was 20-25 cm. 
above the surface of the soil. Except where otherwise specified, 
the readings were made weekly by filling the bottles from a gradu- 
ated burette to a file scratch on the neck. The small area of the 
water surface at this point made the probable error in readings 
less than +0.5 cc., and this could have had no appreciable effect 
upon the results. The instruments were all standardized to the 
same unit before being used, restandardized at intervals of 6-8 
weeks during the season, and a final correction made on their being 
collected in the autumn. By the coefficients thus obtained all 
readings were reduced to the standard adopted by LivincsTon (5) in 
his recent paper on the operation of the porous-cup atmometer. The 
directions given in that article were so closely followed that it is 
unnecessary to detail further the methods used in operating the 
instruments. Two or three atmometers were discarded during the 
season on account of various irregularities in their operation, but 
others either maintained a uniform rate of water-loss or showed a 
variation that progressed uniformly at a readily calculable rate. To 
provide still further against the possibility of serious error, two 
instruments were often maintained a few feet apart at the same 
station, and several stations were usually established in the same 
association, the mean of the various readings being taken as giving 
the true measure of the evaporating power of the air for that asso- 
ciation. | 
No correction has been made for errors caused by rainfall, 
although during showers some water undoubtedly passes through 
the porous cup and into the reservoir, because it was thought that 
the amount of variation thus produced would be the same for all 
stations within so limited an area, and hence the comparative rela- 
tion of results would remain unchanged. This assumption has been 
largely verified by Brown (6), using an atmometer with a rain- 
