1896.] NEW YORK ACADEMY OF SCIENCES. 5 



H, a thermometer fastened to the support A', which is used in 

 dew-point determinations. 



It was important that the scale of percentages, G, Fig. II., 

 should be carefull}^ tested and either verified or the corrections 

 obtained throughout; therefore observations were made with all 

 possible different percentages of saturation of the surrounding 

 air. 



The readings of the hair hygrometers in each case were com- 

 pared with the relative humidity obtained from observations 

 made with wet and dr}^ bulb thermometers, and the percentage 

 of saturation deduced from the Smithsonian hygrometrical tables 

 of Gu3'ot. This was done as a relative comparison, since the 

 wet and dry bulb thermometers, or Auguste's psychrometer, is 

 the means almost entirely used at the U. S. Weather Bureau 

 stations for obtaining the relative humidity of the air. 



Of course, in the present investigation, the detei'minations 

 made with the wet and dr}- bulb thermometers were themselves 

 subject to some error ; yet this method is so generall^^ accepted, 

 and is the means which is so often used for obtaining the rela- 

 tive humidity, that it seemed fair to compare the readings of the 

 hair hygrometer with those calculated from observations made 

 with wet and dry bulb thermometers. 



The results of the comparisons which were obtained indicated 

 that for the middle section of the scale of the instrument under 

 consideration, say from 20 to 85 per cent., only a difference of 

 from 1 to 3 per cent, could be observed. 



But for the extremities of the scale, from to about 20 per 

 cent, and from about 85 to 100 per cent., the reading indicating 

 the relative humidity seemed unreliable, and especially so at low 

 humidities, differing in some cases as much as 10 per cent, from 

 the calculated degree of saturation. 



Thus, in cases of either very low or very high humidity, when 

 two or more hair h3^grometers were placed in the same atmos- 

 phere, their readings were very apt to indicate different relative 

 humidities, and also when the same hair hygrometei was placed 

 at different times in an atmosphere of a constant hygrometric 

 state (of either very low or very high relative humidity) it gave 

 different percentages. 



These variations of course presented a difficult}' in drawing a 

 correction curve for the extremities of the scale on the hair 

 hygrometer. 



Prof. Rood called my attention to an article in the BeAhldtter 

 zu den Annalen der Physik und Chemie, Vol. XIX., No. 11, 

 page 875 : '' Theorie des Haarhygrometers, by B. Sresnevsky," 

 in which it is stated that the change in the length of the hairs 



