278 PROCEEDINGS OF THE AMERICAN ACADEMY. 



merits of the actual slight contractions or expansions which take place on 

 mixing gases seem to have been made. These are, however, so small, 

 that for ordinary purposes they may be neglected. 



It may be a matter of interest to call attention one by one to the 

 percentage effect of the various probable sources of error in the deter- 

 mination of oxygen in air, supposing in each case that the particular 

 error under consideration is the only one present. The analysis is sup- 

 posed to be conducted in a room having the atmospheric conditions of 

 20° temperature and 760 mm. pressure, with the apparatus described 

 above. The error is recorded in percentage of the total volume taken. 



(1) Omission of the correction for aqueous vapor 0.4 per cent. 



(2) Error of 1 mm. in reading water column . . 0.01 " 



(3) Constant error of 1° in thermometer in water- 



jacket 0.02 



(4) Change of 1° in temperature of water-jacket 0.5 



(5) Constant error of 1 mm. in reading barometer 



column 0.02 



(6) Change of 1 mm. in atmospheric pressure . . 0.13 



(7) Admission of absorbing liquid into measuring 



( from ... 

 bulb, according to amount 



.. 



to perhaps . 1 " 



(8) Maximum error from adhering water, perhaps 0.5 " 



(9) Possible error from adiabatic contraction, per- 



haps 1 " 



(10) The retention of a bubble of gas 1 cm. long 



in the 1 mm. capillary 0.02 " 



That all the significant errors may be sufficiently avoided by reason- 

 ably careful work is clear from the actual results ; but the enumeration 

 of their widely varying effects may be useful in showing the novice 

 where to employ his precautions. Most of these errors apply equally 

 to any kind of gas-analytical work, but many of them are persistently 

 disregarded in common practice. 



If the mixture to be analyzed contains gases easily soluble in water, 

 mercury must of course be employed in the measuring bulb, and a good 

 cathetometer should be used for measuring the changes of pressure. A 

 poor cathetometer is often worse than none. 



