OF ARTS AND SCIENCES. 9 



be anticiiiatecl, because of changing curvature of the coils owing to 

 expansion or change of position. Several straight capillaries were 

 also studied, varying in size and in the form of tube employed for 

 heating the gas. These experiments covered the whole 126 meas- 

 urements of this series, and resulted in the selection of the form 

 and size given in the general description of the apparatus at page 2. 

 No systematic study was made, however, of the effects of the propor- 

 tions of the tube on the law of the transpiration, though the apparatus 

 seems well adapted for that purpose. 



Fourth Series. October, November, and December, 1878. — The 

 capillaries used in this series were those described in the general ac- 

 count of the apparatus at page 2 ; viz. straight tubes of about 30 cm. 

 in length and 0.0110 cm. in diameter, and were cut from tube / of 

 the apparatus of 1876. The mercury columns were read to 0.1 mm. 

 by a reading telescope from steel millimeter scales placed behind 

 the gauges. Corrections were applied to these scales, but the pre- 

 cision of reading was much less than in the fifth series ; the tempera- 

 ture measurements were also somewhat less precise, and the whole 

 series should receive much less weight, owing to the conditions under 

 which it was taken, than the fifth series on air, or the second series on 

 carbonic acid. Table III gives the results obtained. 



In the Table the letters heading the columns have the same mean- 

 ing as elsewhere throughout the paper. The pressures jo^, p^., &c. 

 are given in millimeters of mercury at 0° C. at Boston. The column 



headed " Mean — or K." contains the mean of the s^roup of values of 



— , or of K, immediately preceding, and a similar statement is true 



'7i 



of the column headed " Mean 4-" 



The results from No. 49 onwards are of much greater precision 

 than those preceding, chiefly from the use of greater pressures, 

 which increased the precision of measurement. Measurements 49 to 

 99, when combined by the method of least squares, give for constants 

 in the empirical equation 



^'=^2 = 1+^^ + ^^' 



% '?! 



the values, between the limits f = 0° and t = 100°, 



^r= 0.002821, and 5 = —0.00000149. 



These values, however, I have regarded as entitled to so much less 

 weight than those of the fifth series for air, that they are given here 



