21 S PENDULUM OF COMPENSATION. 



to think these speculations the mere creatures of theory; nnci 

 may, from the actual performance of many surprising time- 

 measurers, be disposed to conclude, that they may be practically 



Di'^quisitions ^f ,-jq consequence. But I would urse. Sir, that it is by this 

 upod minute .... o ' ' j 



changes in kind oi' minute and scrupulous investigation, that discoveries 

 bodies useful, come to be made ; and that, while the best makers hold different 

 opinions, and with all their skill (which I truly honour) cannot 

 foretel, beyond a certain point, whether the work they are 

 upon will answer their expectation or not,— we may actually 

 expect still to make discoveries^in this useful and curious pur- 

 suit. And while we make our experiments and operations 

 under the guidance of some probable hypothesis, or doubt, or 

 ^ indication, the objects we produce may, in many instances, 

 be found to possess unforeseen advantages, or serve to establish 

 new truths, perhaps not suspected by the operator himself. 

 New pendu- I have been desirous of constructing a pendulum, composed 

 of expansion bars, not subject to the violence of soldered or 

 adherent faces pulling against each other, or the uncertainty 

 of contacts, the rub of surfaces, the stickage of joints, or the 

 bearing parts of levers. The following has been going for 

 seveial years. 

 Construction. In the annexed drawing, (Fig. 2, plate v.) f g and i h repre- 

 Two pair of g^f^j- }q^^.^ ^f bi-^cg respectively united by rivetting and 



strait bars, ,, . . , ^_ , , , , , 



each of brass soldermg With two bars of steel at the ends where they touch, 



without, and but their faces do not touch except near the ends, as shown in 

 steel within, i i - 

 are rivetted ^^e drawing. 



tog-ether, and ^t the extremity g 1 n i of the system, all the four bars are 

 the other at Connected, as is also shown in the cross section p ; but at the 

 one end only, other extremity the bars f k are connected, and so likewise are 

 ed^at^certain^" '^^^ ^^''^ m h 3 but there is no union between k and m. At o, o, 

 thin parts, <S:c. the bars are filed away or notched, so as to leave each of them 

 thin on the side farthest from the middle line of the system, 

 a d shov.'s part of the pendulum rod, of which a is the point 

 of suspension, connected by a spring with b, a clamp adjust- 

 able by sliding along the face of f g, and fixable in any required 

 position by the screw c. And d is pait of the lower rod, broken 

 off to save room in the plate instead of being continued down 

 to the ball. 

 Particular de- The effect of this combination maybe thus explained : When 

 9criptiono t le j^^ ^^ increase of temperature the brass bars fg and i h become 



longer 



