VOL. LXXXV.] VHILOSOVHICAL TRANSACTIONS. 6l5 



chain, which Mr. Ramsden had made for the r. s. For this purpose a distance 

 of 1000 feet was carefully measured with the rods and the chain. The result of 

 these measurements appeared to be such as would have produced a dift'erence of 

 httle more than half an inch on the whole base, had it been measured with each 

 of them respectively. But notwithstanding the apparent degree of accuracy 

 which this, or any other mode of measuring may be supposed capable of, yet it 

 seems necessary that every base, intended to become the ground-work of such 

 nice operations, ought always, when circumstances will permit, to be measured 

 twice at least. 



The manner in which the glass rods were applied in the measurement, is sup- 

 posed to have rendered the operation liable to some small errors, which lying 

 different ways, might possibly have counterbalanced each other, and produced a 

 true result: but this supposition ought never to be admitted in experimental in- 

 quiries, unless such errors can be nearly estimated. The principal cause of 

 error is supposed to arise from the ends of the 2 adjacent rods being made to 

 rest on the same tressel ; because when the first rod is taken oftj the face of the 

 first tressel, being then pressed by the end of one rod only, will acquire a ten- 

 dency to incline a little forward. The error arising from this cause will evidently 

 tend to shorten the apparent base. Another source of error is supposed to arise 

 from the casual deviation of the rods from a right line, in the direction of the 

 base, tending to increase its apparent length. And a third error is supposed to 

 result from the method which was used, of supporting the ends of the rods on 

 2 tressels only, by which they become liable to bend in the middle. This con- 

 cave form of the rods would also tend to lengthen the base. The first of these 

 causes of error was submitted to experimental inquiry in the garden of Richmond 

 house, Whitehall, in the presence of his Grace the Duke of Richmond, Sir 

 Joseph Banks, Mr. Ramsden, andMr. Dalby; when it appeared evidently, that 

 the glass rod had a small motion when the other rod, which had counterbalanced 

 it, was taken from the tressels. 



These considerations therefore rendered it necessary to compare the measure- 

 ment with the glass rods, with that performed by some other method; not on 

 account of any doubt being entertained of the care with which General Roy's 

 operation had been performed, but solely with a view to bring this new mode of 

 measuring to some proper test. No method of comparison could perhaps be 

 better than measuring the same base with the steel chain. General Roy him- 

 self, in his remarks on the comparative accuracy of the 2 bases, that of Houns- 

 low Heath and Romney Marsh, evidently gives the preference to the chain; 

 which, every circumstance considered, it is certainly right to do. These reasons 

 induced the Duke of Richmond to direct the base on Hounslow Heath to be re- 



