562 SCIENCE PROGRESS 



accurate means of estimation. Fortunately, such a method has 

 been introduced by Waller and Geets, and perfected by the 

 former author. Levy (5 d) has shown that this method gives 

 practically identical results with that of the combustion method of 

 Vernon-Harcourt (4^). Careful experiments by Waller's method 

 showed that under the conditions of the present research the 

 weights could be relied upon to within 1 milligramme — that is, 

 the error of each estimation was less that "i per cent, of 

 chloroform. 



Greater accuracy than this requires that rather elaborate 

 precautions should be taken, so the arbitrary rule was made 

 that the graduations of the machine should be correct only 

 within this limit of error. It was found subsequently that a 

 variation of less than '25 per cent, was quite inappreciable under 

 the ordinary conditions of human anaesthesia, and so for the 

 present purpose the limit of error of 'i per cent, gives quite 

 sufficient accuracy. 



A scale was then constructed, and was tested under all the 

 conditions which could be expected to arise. It was found that 

 many circumstances could be neglected in practice. For 

 example, the level of chloroform has only a very slight influence 

 on the percentage, so that, even after working twenty minutes, 

 the percentage was practically unchanged. If the apparatus is 

 accidentally shaken, this produces no change in the percentage. 

 A small amount of water condensed from the air has an effect 

 only within the limits of error, and variation in the air supply 

 between 8 litres and 20 litres per minute is also without effect. 

 This last result is of importance and was specially examined. 

 Levy has already referred to the physical law on which this 

 depends (1). 



The effect of temperature remains to be considered. As the 

 temperature rises more chloroform evaporates, and the per- 

 centage at the same opening increases. This temperature 

 correction is greater with a high percentage than with a low. 

 But the construction of the sectors is such that a given length of 

 movement causes a greater increase in the percentage when the 

 sectors are nearly open than when they are nearly shut, so that 

 the graduations of the scale marking the percentages are closer 

 together at the high values than at the low. 



If, then, such a scale is made to slide bodily to right 

 or left in accordance with the temperature, a given move- 



