Methods of Gas-analysis. * , «<-Qf_ihese the most important by far in 

 Plant Physiology apply to carbon dioxide and oxygen. Somewhat elaborate 

 but very exact methods have been described by Hempel 

 and Winkler (their books: Darwin and Acton, 6; 

 MacDougal, 235), by Timiriazeff (his micro-eudiom- 

 eter, Darwin and Acton, 45), by Ponnier and Mangin 

 (MacDougal, 258, though the accuracy of this apparatus 

 has been questioned by Pantanelli in a paper cited under 

 Literature). I have mysel f devised a method, much 

 simpler than any of the above, but amply efficient for all 

 except precision work. It makes use of the usual 

 absorbing chemicals described elsewhere (pages 98 and 

 103), but applies them by means of reagent tubes 

 illustrated in the accompanying figure (Fig. 25). The 

 gas to be analyzed is collected or brought into a gradu- 

 ated tube, — over water if the absorptiQn by the water 

 .is negligible, otherwise oyer mercury. Then a reagent 

 tube of the same diameter, sealed at one end and provided 

 at the other with an extension of stout rubber tubing, is 

 filled to near the top of the rubber with either the caustic 

 potash solution, or the potassium pyrogallate, as need 

 may be, and is sealed by a screw clamp as in the figure. 

 It is then placed under the water (or mercury), all air 

 is carefully squeezed from the tube above the clamp, and 

 the rubber is slipped over the open end of the graduated 

 tube, which it should grip firmly. The whole is then 

 lifted from the water, the clamp is opened, the combina- 

 tion is inverted, and the liquid is allowed to flow back 

 and forth several times from one tube to the other, when 

 it will completely absorb any carbon dioxide (or oxygen) 

 present. The combination is then held upright until the 

 liquid has all settled downward, when the clamp is 

 closed. Then the combination is slipped again under 

 water, and the rubber tube is pulled off, when the atmos- 

 pheric pressure will instantly force up- the water to the 

 exact extent of the gas absorbed,, permitting the amount 

 to be read off directly. The corrections for capillarity 

 and vapor-tension may be ignored in demonstration, 

 though in exact work they would be taken into account, -p „ , , . .,,,.,, 

 Barometric pressure obviously can introduce no error, /' '„ \ „„„„ji. 



1 J . ' ^LOWERJ AND GRA- 



but temperature changes must be compensated either by 



calculation or by reading the graduated tube both before 



and after the test, while a stream of water of constant 



temperature is flowing over. it. A source of error to be 



guarded against is the possible leakage of air into the 



tubes as the gas is absorbed and the internal pressure reduced. This can 



be prevented by tight-fitting tubing, especially if the ends be tightly wrapped 



DUATED tubes, 

 with rubber 

 connection and 

 clamp; Xf. 



