284 E. W. Morley — Volumetric Composition of Water. 



volumes as nearly equal as possible, and then to have measured 

 a small excess at the point at which the excess was to be 

 measured after the explosion. But this was not thought of in 

 time As it was, errors of calibration could have no great 

 effect ; but they were further made of still less effect on the 

 final mean in two ways. In some experiments, the hydrogen 

 was put in excess, and in some, oxygen ; if in two such experi- 

 ments the sum of the two gases were the same, the points of 

 measurement of oxygen and of hydrogen would be inter- 

 changed, and the errors of calibration would produce contrary 

 effects. Again, the amounts of gas taken were increased from 

 time to time, so that all points from 63 to 75 centimeters 

 were used, hj which also accidental errors in calibration were 

 rendered of small influence on the result. 



The measured gases were now ready for explosions in frac- 

 tions. Explosions were always made in the presence of a large 

 volume of inert gas. The ratio of explosive gas to inert was 

 varied within somewhat wide limits, and the same ratio was 

 preserved throughout all the explosions of a given experiment. 

 Suppose that in a given case the ratio desired was that of four 

 to one. To the 120 cubic centimeters of hydrogen left in the 

 eudiometer were added 30 from one of the jars containing 

 hydrogen and oxygen. After this was exploded, a like volume 

 was added again, and so on, till all the gas previously mixed 

 had been consumed. There would still be 120 cubic centi- 

 meters of hydrogen in the eudiometer. To this was now added 

 eight cubic centimeters of oxygen from the small graduated 

 jar. The two were mixed by letting mercury drop through 

 the eudiometer, and were exploded. A smaller computed 

 volume of oxygen was added and mixed for the next explosion, 

 and so on, till all the oxygen was finally exploded in presence 

 of fifteen times its volume of hydrogen. But when oxygen 

 was to be finally in excess, a variation was made a little before 

 the last explosion, by adding small quantities of hydrogen to 

 an excess of oxygen in the eudiometer ; up to this point, hydro- 

 gen was kept in excess. It is obvious that the last explosion 

 of the series is the critical one on whose completeness accuracy 

 depends ; sufficient attention was given to this matter. 



After the explosion was completed, it cost some trouble to 

 put the eudiometer into condition for good measurement, 

 because so much water had accumulated in it. Then an excess 

 of oxygen or of hydrogen, as the case might be, was added, the 

 mixture was exploded and the residue was measured. From 

 this was computed the amount of nitrogen in both gases taken 

 together ; subtracting the nitrogen known to exist in the 

 hydrogen according to the previous experiment the remainder 

 was the nitrogen in the oxygen used. The ratio of the volumes 



