AND PLANTS 89 



earnest effort to reduce the margin of experimental 

 uncertainty led to an important and quite unexpected 

 discovery. 



A dozen years ago, Lord Rayleigh had a certain glass 

 flask, which he weighed and measured very carefully, by 

 means I cannot stop to describe. When he had done 

 this, he weighed the quantity of oxygen this flask would 

 hold at a temperature of 1 5C. and a pressure correspond- 

 ing to a barometric height of 760 mm. of mercury. Here 

 are his final results. You see that if he had only kept 

 four figures in his records, they would be in absolute 

 agreement : but the fifth figure begins to show slight 

 discrepancies. 



Weight of Oxygen in a Flask at 15 C. and 760 mm. (Rayhigh\ 



2-6272 grammes 

 2-6271 

 2-6269 

 2-6269 

 2-6271 



In these records, each unit of the fifth figure means 

 a ten-thousandth part of a gramme, which is a weight too 

 small for any of us to perceive with our unaided senses ; 

 it can only be detected by a skilful observer provided 

 with a delicate balance. The differences between these 

 results are therefore very small indeed ; but the whole 

 quantity measured is also small, and the range over which 

 uncertainty is shown to extend is two or three parts in 

 20,000, a very much larger fraction of the whole result 

 than is affected when we measure an angle of latitude, so 

 that the accuracy attainable in the one case is much 

 greater than in the other. This is due to the very com- 

 plicated nature of the process by which oxygen must be 



