Page:Philosophical Transactions of the Royal Society A - Volume 184.djvu/579

566 we get for Neglecting altogether Series I., as in all probablityprobability [sic] (as pointed out above) the oxygen was contaminated by chlorine, we have the values—

as the extreme values for the number of volumes of hydrogen uniting with 1 volume of oxygen to form water. As Series D was a small one, especially when the two first experiments in it are omitted, the second value is unduly weighted by the extremely small probable error in the remaining three, I think that the value—

may be taken as the true value for the ratio sought.

It is highly satisfactory to note that this is exactly the value found in the last series of twelve experiments made with the purest gases, and the advantage of prolonged experience in the manipulation of these difficult and tedious experiments. Although the last of the experiments was performed in April of last year, I have refrained from publishing the results till I could feel certain that I was unable to detect any further source of error, and until I felt assured that all the possible sources of error which have been suggested to me by various friends who have taken an active interest in my work, had been successfully overcome or did not apply to my methods of manipulation and measurement.

With regard to the results published by, I think I may fairly claim that the simplicity of my apparatus for measuring and preparing my gases, the fact that they were never transferred from one mercury trough to another, that they were always saturated with water in the same way, and that the water formed by the