550 " CAPILLARY ACTION. 



curve, or are within the limits of the angle of repose on eirr 

 ther side of the perpendicular. But if the wall is very low, 

 and the arch flat, a segment of a circlp will be more cor- 

 rect than a parabola. Hence it is obvious, first, that a seij;- 

 Circle prefera- ment of a circle is a better form for an arch than an ellipsis 

 lip^sis° " ^^ P^iual heij^ht and span, although less pleasing to the e^e^ 



the horizontal thrust being less: secondly, that for the same 

 Pointed arch reason, a Gothic or pointed arch is preferable to a Saxon or 

 rases"^ '^^^^^^^ semicircular arch, when its heiglit is greater; and even when 

 the height is equal, an arch composed of two panibolic seg- 

 ments meeting in the vertex is stronger than a semicircular 

 arch : for, supposing the whU very high, the depth of the 

 arch stones of a semicircular a^ch must be at least yV of 

 the span, in order that the arch may stand,, but that of the 

 stones of a Gothic arch, composed of two parabolic seg- 

 ments, may be less by one twentieth ; the parabola of equi- 

 librium touching in this case the internal limit of the arch 

 In ptliers the at -j-Vir of its whole height above the abutments. If, how- 

 circular. ^\^x^ the arch is flatter, a segment of a circle will be some-r 

 what stronger than a pointed arch composed of parabolic or 

 elliptical segments. When the arch is higher, it is obvious 

 that a single circular curve is no longer applicable : and in 

 this case, it is of little consequence whether the segments 

 be circular or paraboSic, either of these forms approaching 

 sufficiently near to the curve of equilibrium, and both pro- 

 ducing equally a much smaller horizontal thrust than a se- 

 micircular arch. 



TI. 



Additional Remarks on the capillary Actions of Fluids. By 

 Aletes. 



To Mr. NICHOLSON. 

 SIR, 



Capillary ac- JL T has been observed, with apparent justice, by Mr. 

 Laplace, that the force of capillary action, other things 

 being equal, must be proportional to the square of the den- 

 sity of a liquid; and it is easy to deduce this result from 



the 



