24 BULLETIN OF THE 



When the preliminary experiment had proved that the air which for 

 twenty-four hours was being pumped through the apparatus did not 

 contain enough carbon dioxide to cause any trace of a precipitate in the 

 baryta bulb, then the tubing was clamped at x. Equilibrium was re- 

 stored by forcing in air as before, four fishes were introduced through the 

 neck in the cover to the jar, and the pumping was resumed. It was kept 

 up for six hours, at the end of which time not the slightest trace of a 

 precipitate had been formed, although the fishes had been emitting bub- 

 bles at the surface of the water in the ordinary manner during the whole 

 time. The baryta water was subsequently tested to make sure that no 

 error had been made in the matter of its sensitiveness to carbon dioxide. 



The immediate inference from this experiment is, that the gas elimi- 

 nated by the fishes contained no trace of carbon dioxide; but here, again, 

 the possibility of an absorption of eliminated carbon dioxide by the water 

 cannot be rigidly excluded. 1 It seems highly improbable, however, that 

 with an air-chamber which required at most only a few minutes to be 

 swept of its contents, there could have been a sufficiently prolonged 

 exposure of the gas to allow the absorption of its carbon dioxide. It 

 might be urged that this, being but a single experiment, can be entitled to 

 little weight, — that, even if in this one instance there was no elimina- 

 tion of carbon dioxide, it is to be considered simply as an exceptional in- 

 dividual 'nse. It is, however, to be borne in mind that there were four 

 fishes under experimentation, and that in the case of the analysis which 

 failed to show the presence of carbon dioxide there were six. 



As the result, then, of these various experiments to determine 

 whether carbon dioxide is eliminated, it must be concluded for the 

 present, until opportunity is^ presented for a careful analysis of the 

 gas taken from the bladder without exposure to water, that the gas 

 of the air-bladder of Lepidosteus is not likely to contain more than two per 

 cent of carbon dioxide at the most, ivith a strong probability that in many 

 cases it does not contain an appreciable amount of that gas} 



1 If this experiment were ever repeated, it is perhaps desirable that control 

 experiments should be carried on at the same time. In the control experiments 

 atmospheric air mixed with known volumes of carbon dioxide, varying from zero 

 to 2% of the total volume, should be made to bubble near the surface of the water. 

 By observing the frequency and estimating the volume of the gas emitted by the 

 fishes in a given time during the actual experiment, practically the same conditions 

 could be observed in the control experiments. The introduction of the carbon 

 dioxide mixture could easily be accomplished by means of a third glass tube, 

 piercing the rubber stopper of the jar and dipping just below the surface of the 

 water. 



2 The recent analyses made by Jobert ('77 and '78 a ) on the gases taken from 



