Death in Closed Vessels 523 



Again I make a supposition: the present level will be, let us say, 

 between 20.2 cc. and 20.3 cc. and I must choose between 20.23 cc. 

 and 20.28 cc. According to whether I take my meniscus level 

 higher or lower — and we know how difficult the estimate is for a 

 colorless liquid, without mentioning the fact that the meniscus 

 of pure water in a tube which is necessarily rather dirty is not 

 the same as the meniscus of a potash solution which wets the glass 

 perfectly, — there will have disappeared in an average quantity 

 of 25.35 cc. either 5.04 cc. or 5.14 cc. of carbonic acid, which gives 

 for the percentage composition in the first case 19.88 cc. per 100 

 (25.35 : 100 = 5.04 cc. : x = 19.88) ; and in the second case, 20.27 cc. 

 per 100 (25.35 : 100 = 5.14 cc. : x = 20.27) . 



That is, without taking into account the cause of error due to 

 the operation itself, which here involves the second decimal, my 

 analysis, though made as well as possible by the volumetric method, 

 exposes me to an error which in this present , case may equal 

 0.39 cc. In other words, the first decimal can and must be con- 

 stantly vitiated in the necessary speed of the analyses. 



It is with this consideration that we must examine all the 

 results of our analyses; and therefore no one will reproach me 

 for having prudently stopped at this figure which indicates exactly 

 the degree of precision which one can expect from this volumetric 

 method of analysis, any more than for not having used other 

 methods, when the experimental differences, over which we have 

 no control, are at least of the same order, as I said above. 



With these reservations, let us examine the results of the ex- 

 periments which, in order to shorten tiresome reading, I have 

 grouped in a summarizing table, arranging them in the order of 

 the pressures. 



If we examine Column 8, which gives the proportion of oxygen 

 remaining in the air which has become irrespirable, we see that 

 the numbers which it contains increase proportionately as the 

 pressure diminishes. This rule, however, has apparent exceptions, 

 as the table shows. But if we consider the results given by the 

 analyses of air which has become incapable of supporting life 

 made at the same pressure, we see that the exceptions mentioned 

 are of the same order as the differences which separated the re- 

 sults of these analyses. Thus at normal pressure, the proportion 

 of oxygen remaining varied between 3.0 and 4.2 per cent; similarly 

 at 24.2 cm., it varies between 11.6 and 13.7 per cent. 



The general tendency of the phenomenon is shown still more 

 clearly in the curve represented by O, Figure 17. Here the pres- 



