ON THE RESPIRATORY FUNCTION OF STOMATA. 21 



the absorption chambers. When titrations are being made, 

 and the air current can no longer be allowed to pass through 

 the absorption chambers, it passes through a column of 

 water equal in its resistance to that of the baryta solution in 

 the absorption chamber. This enables the rate of flow to 

 be kept constant between, as well as during, the actual 

 experiments. Numerous other details, such as the special 

 method of refilling the burettes, etc., and above all those 

 small points by which constancy is, as far as possible, attained, 

 many of which have involved weeks of special experiment, 

 cannot be described here. 



Simplifications of technique by complication of apparatus 

 has been the guiding principle, and the result is that, 

 although the whole consists of at least eight separate pieces 

 of apparatus, many being further in duplicate, and all con- 

 nected together by a plexus of tubes, yet the working is so 

 automatically arranged that the operator, beyond reading the 

 burettes and occasionally working a finger bellows, has no- 

 thing to do but turn stopcocks. 



If only one series of estimations is being made, these can 

 be kept absolutely consecutive, the current being led through 

 one of the absorption chambers, while the solution in the 

 other one is being titrated and renewed, and so on alternately. 

 When two series of comparative estimations are being made 

 at once, a small interval must be allowed after each double 

 estimation, during which the titrations are performed ; the 

 currents of air in connection with the plants then pass 

 through by-paths, still at their previous rate. 



Delicacy of estimation sufficient for present work is ob- 

 tained bythe use of half-decinormal, N/20, standard solutions. 

 Phenolphthalein is used as indicator, and specially delicate 

 end-reactions can be obtained, since atmospheric CO, is 

 excluded, and moreover the burettes containing both the 

 solutions can be drawn upon. 



Series of control titrations often give an error of observa- 

 tion not exceeding 1/200 c.c. of C0 2 , and 1/50 c.c. is 

 sufficient for a trustworthy estimation in experiments of 

 short duration. 



