EXPERIMENTAL METHODS. 



19 



twice, so that there was an adequate check on any errors. A considerable 

 number of supplementary analyses were made by the writer and the surpris- 

 ingly close agreement gave evidence that individual manipulation does not 

 cut as large a figure in the use of the Bonnier apparatus as might be supposed, 

 and that the accuracy of the results was something more than merely relative. 



By this method the amount of carbon dioxide is obtained by the diminution 

 in volume of the gas sample following exposure to a solution of potassium 

 hydroxide, and the oxygen is determined by the further absorption which fol- 

 lows the addition of pyrogallic acid. Whatever remains of the sample is 

 taken as nitrogen. If nitrogen has been neither absorbed nor given off from 

 the tissues, this remainder will amount to 79.1 per cent, i. e., the quantity 

 present in normal air. If this is the case, the calculation of the percentage 

 composition of the gas sample may be readily made directly from the figures 

 obtained. If on the other hand the amount of nitrogen is more or less than 

 79.1 per cent the following correction must be applied:" 



Let the amounts of carbon dioxide, oxygen, and nitrogen found be respec- 

 tively a, 6, and c. Further let the product of the normal percentage of nitrogen 



in air into that of oxvgen, i. e., ^cr\, be q. Then the actual amount of oxygen 



"^i"i ' *-* ' * 



TABLE 1. 



/ - 



absorbed is eg b, and the Q^ qutotient may be expressed as ~^' 



The question also arose as to whether a correction might not be necessary, 

 for the reason that the experiments were performed and the samples gathered 

 in Tucson, Arizona, at one temperature 

 and pressure, while the gas-analyses were 

 mostly made in New York at another 

 temperature and pressure. Investiga- 

 tion showed, however, that the correc- 

 tion due to the nonconformity of carbon 

 dioxide to the gas law would affect results 

 less than 0.1 per cent, which is below the 

 limit of experimental error. 6 Conse- 

 quently this correction was disregarded. 



In comparing the results obtained by 

 this method with those from the Pet- 

 tenkofer apparatus, where the carbon 



dioxide is estimated by weight, it is necessary to reduce them from cubic 

 centimeters to milligrams. As most of the experiments were carried on at 

 Tucson the barometric pressure of that locality must be taken into account. 

 The average fluctuation is from 690 to 700 mm. and the calculations were 

 made on that basis. Table 1 was compiled from figures obtained by extrapo- 

 lation of known data c and is the weight in milligrams of 1 c.c. of carbon 

 dioxide, saturated with water-vapor at the different temperatures and pres- 

 sures indicated. 



"Aubert, M. Recherches physiologiques sur les plantea grasses. 2d part, p. 23, Paris, 1892. 

 Pallidin, W., and S. Kostytschew. Methoden z. Bestimmung d. Athmung d. Pflanzen. la 

 Abderhalden's, Handbuch der biochemischen Arbeitsmethoden, vol. HI, part 1, p. 500. 



ft Amagat. Ann., Chim. Phys., series 4, vol. 29, p. 246, 1873. F. Fuchs. Wiedermann'a 

 Annalen, vol. 38, p. 302, 1889. 



c Castell-Evans, J. Physicochemical tables, vol. 1, pp. 354-357, table LIX, J. London. 1902. 

 Calculated from the data of Regnault. 



