18 MF/rHOD^ OF ISVE^TlfiATlON 



that it is really oxygen-free. This is not the case with 

 the commercially produced gases. Bubbling them through 

 one or two gas wash-bottles containing alkaline pyrogal- 

 late, as has sometimes been done, is not sufficient to re- 

 move all the oxygen. The gases should V)e led, before 

 reaching the animal chamber, over copper or platinum 

 that is sufficiently heated. After the chamber has been 

 flushed by a rapid stream of purified gas, one can main- 

 tain a slow stream during the rest of the experiment. If 

 the latter is of long duration, however, this is not always 

 convenient. One may then create a higher pressure in- 

 side the chamber than prevails outside, so that, after 

 outlet and inlet have been closed, the gas would leak 

 out and no air would enter the container. But even then 

 it is probably safest to pass a fresh stream of gas 

 through the chamber from time to time. In all exper- 

 iments of this type it should be remembered that rubber 

 is not impermeable to gases. The ideal is to dispense 

 with all rubber connections after the gas has been passed 

 over the heated copper or platinum. If this cannot be 

 done, the rubber connections should at least be reduced 

 to a minimum. 



To replace completely the oxygen dissolved in water 

 by hydrogen or nitrogen is more difficult. It is clear that 

 mere boiling of the water is not sufficient to remove all 

 oxygen. Alsterberg (1922) found that even after long 

 boiling and care to insure that during the cooling process 

 only a small surface was exposed to the air, about 0.1 

 cc. oxygen per liter remained. The oxygen tension can 

 be reduced further by bubbling oxygen-free hydrogen or 

 nitrogen through the water after the latter has been 

 boiled. In this way one can easily arrive at solutions that 

 are sufficiently oxygen-free to be satisfactory for most 

 experiments on larger aquatic animals. 



The closing of the experimental vessels should, need- 

 less to say, be such that they are absolutely airtight or 

 else a constant stream of gas should be maintained. The 



