224 Proceedings of the Royal Society of Victoria. 



convenience and indeed accuracy, and as Colonel Herschel has 

 proposed, to use the vacuum apparatus only for ensuring a 

 constant pressure of say twenty-seven or twenty-eight inches, 

 except of course for the one set of swings at low pressure taken 

 at the station of reference, However, much would depend on 

 the condition of the vacuum apparatus. 



I will mention here lest I should forget it that it is well to 

 allow an observation (whether by a single swing, as may be done 

 in vacuo or by a succession of swings does not much matter) to 

 extend over twenty-four hours, or if that he inconvenient at 

 least from dark to dark, through day or night as may be chosen, 

 so as to rate the clock by transits for the interval of time over 

 which the observations extend. For you cannot trust a clock, 

 even though the rate from day to day be very uniform, to be 

 quite exempt from a diurnal inequality of rate. 



(b) Suppose now that we prefer to depend on experiment for 

 the correction for the air. Then we may choose our form of 

 pendulum as we please. That usually employed has the bar 

 somewhat thin, in a fore and aft direction, so as to be slightly 

 flexible. Without this there is, I believe, some difficulty in 

 ensuring that the weight shall bear well on both agate planes, so 

 as not to run the risk of turning slightly about a vertical axis to 

 and fro as it swings. I recollect someone (Sabine, I think), 

 telling me that someone, I forget who, did not like the flexibility, 

 and proposed to make the pendulum stifi", and Ivater (I think it 

 \yas) said, " He'll find it will not do." 



The form having been chosen, we have to find the correction for 

 the air experimentally. This demands the use of a vacuum appara- 

 tus. I think the most convenient plan would be to get a. facsimile 

 of the pendulum made of wood. The resistance of the air depends 

 only on the form and time of vibration of the pendulum, I mean 

 supposing the state of the air given, and these would be the same 

 for the actual pendulum &nd for the wooden model. By avoiding 

 a specially dense wood we might easily get the model ten or 

 twelve times as light as the actual pendulum, and the eflfect of 

 the air on arc and time would be magnified ten or twelve times. 

 The whole time of the swing would be reduced in the same 

 proportion ; but this would not signify as regards having a 

 shorter interval by which to divide any error of observation of 



