548 Prof. Forbes's Researches on Heat* Secoiid Series. 



The mode of observation by the first impulsive arc I have 

 invariably adopted for obtaining numerical results, and chiefly 

 for these reasons: 1. It saves time, and thus renders conse- 

 cutive observations comparable. 2. It prevents a long ex- 

 posure of the pile to heat, which alters the zero point and in- 

 jures its action. 3. It almost annihilates the effect of con- 

 duction where substances, capable of retaining heat, are placed 

 between the source of heat and the pile. 



It is a remarkable circumstance, that when both the cor- 

 rections obtained from these tables are applied, we obtain (as 

 far as 20° at least) a measure of intensity increasing, almost 

 uniformly, with the arc first run through. This is found to 

 depend on the circumstance that the curve, expressive of 

 forces, in terms of the stationary deviation, is convex towards 

 the axis, or the forces increase more rapidly than the arcs ; 

 whilst the curve, expressing the stationary in terms of the 

 momentary deviations, is concave to the axis, or the Statical 

 effect increases in a less ratio than the Dynamical effect. The 

 convexity of the one curve, almost compensating the conca- 

 vity of the other, the relation obtained between the first im- 

 pulsive arc and the calorific force is nearly linear. 



This will be best illustrated by comparing the true ratios 

 of the forces obtained from the above table, with the simple 

 ratio of the first deviations ; and to put it in greater evidence 

 we shall suppose a deviation of 20°, which is greater than we 

 have ever employed in these experiments. 



Since, therefore, even in this case, we should never have an 

 error amounting to a unit in the second decimal place, I have 

 contented myself in this paper with the employment of the 

 simple arithmetical ratio of the first arcs passed over. 



It appears, then, that these effects are developed on the 

 whole in a simple and uniform manner; and though such an 

 investigation as we have undertaken of the instrument, was 

 necessary to give us confidence in the numerical accuracy of 

 the results, iA\ facts of importance might be determined with- 

 out it, and even quantitative laws ascertained by a judicious 

 conduct of experiments. Many persons, even though not 

 unaccustomed to physical reasoning, have strangely inaccu- 

 rate conceptions of the limits of possible errors. Nor is there 

 a more important part of the science of experiment than to 



