522 Experiments 



diploe, in the occipital region. Lived 1 hour 30 minutes in the equiva- 

 lent of 1.6 liters of air, or 56 minutes per liter. 

 Lethal air: O 8.2; CO? 11.6. 



CO. 



CO + O* = 19.8; == 0.91. 



O2 



I have purposely given in the preceding pages an account of a 

 great number of experiments so as to show what is indefinitely 

 variable in the phenomena and at the same time what stands out 

 as general in this variety, which defies both deceptive averages 

 and sham precision of decimals. Certainly, when a sparrow dies 

 under a bell at a certain pressure, the air of this bell has a com- 

 position which the best methods of modern chemistry could per- 

 haps permit us to determine to about one ten-thousandth. But 

 what would be the use of this precision when our experiments show 

 us that another sparrow exactly like the first, placed in apparently 

 identical conditions, dies with a composition of ambient air which 

 may differ from the first by 4 or 5 tenths of oxygen or carbonic 

 acid, or even more? It is evidently better to multiply experiments 

 to try to find the explanation of these differences and to adhere to 

 convenient methods of analysis which permit one to work rapidly. 



But the height of absurdity — and this, unfortunately, is found 

 rather frequently in German work — is to claim to give to these 

 other methods an appearance of precision which they do not pos- 

 sess, carrying calculations to the second and third decimal and 

 even resorting to a table of logarithms to get more decimals. This 

 charlatanism of decimals which leads one to claim exactness for 

 the thousandths in a number which is wrong beyond the units, is 

 an illusion which must be avoided. Let us make our criticism 

 specific by applying it to the present case. 



Let us imagine, in a tube graduated to tenths of cubic centi- 

 meters, inverted over the mercury bowl, our usual gaseous mixture 

 of nitrogen, oxygen, and carbonic acid. To avoid taking account 

 in the first determination of a convex mercurial meniscus and in 

 the other two of a concave aqueous meniscus, I first introduce into 

 the tube some drops of pure water, and try to determine the level. 

 Now admitting that the greatest precautions have been taken, it is 

 impossible to estimate the height of the liquid column with a closer 

 approximation than five hundredths. Let us suppose that I have 

 found that it is between 25.3 cc. and 25.4 cc; but I cannot be sure 

 whether it is 25.32 cc. or 25.37 cc, for example. I now add the 

 potash, shake it vigorously and repeatedly, and again plunge the 

 tube into the mercury to bring it to its original temperature. 



