CAUSE OF CIRCULATION OF HOT WATER. 267 



bend, so as to form a siphon, may rise 

 higher than the feeding cistern c and air- 

 pipe a; but this construction is danger- 

 ous, inasmuch as air may accumulate in 

 the bend and stop the circulation. It is 

 better, therefore, to make the cistern c 

 and air -pipe a higher than any required 

 elevation in the water-pipes, as m, and 

 furnish the latter with a small pipe to 

 carry off the air. Every such ascent and 

 descent evidently creates a force of some 

 amount in favour of the movement, not, 

 perhaps, more than enough to balance the 

 increased friction produced by the bends, 

 because the temperature of the ascending 

 and descending columns cannot differ 

 much. When it is, however, necessary to 

 make a descent, as n, the difference of 

 temperature in the two columns is ad- 

 verse to the movement ; and as this acts 

 in addition to the retardation produced 

 by the changes of direction, such descents 

 should be cautiously made, and only 

 when there is an obvious preponderance 

 of power to carry the water through 

 them. 



" In such a system as that described in 

 the figure, where the vertical pipes may 

 be called from 5 to 10 feet, and where 

 there are no descending bends like n, it 

 is surprising through what an immense 

 length of nearly horizoDtal pipes the cir- 

 culation may be carried. I believe," he 

 says, "that the limits are not known; 

 and, as the friction of many hundred feet 

 of pipe must be considerable, though it is 

 less than is generally supposed, the effect 

 seems to require some more powerful 

 cause than the different weights of the 

 two columns. Various speculations have 

 been made with the view of accounting 

 for this, but I am not," he says, " aware 

 of any that is quite satisfactory, and the 

 following may be as little conclusive. 

 It appears to me, however, that the con- 

 tinual cooling of the water furnishes a 

 material facility towards its movement. 

 Referring to the last fig., (348,) if % be 

 supposed to represent the bulk of a cer- 

 tain portion of water at that place, and y 

 the smaller bulk that it would afterwards 

 occupy in consequence of its loss of heat 

 in the interval, it is clear that there must 

 be a rush to fill up the void thus created ; 

 and as this rush would take place with 

 the greatest facility in the direction of 

 the moving mass, there would exist at 



every part of the circulation an impelling 

 force dependent on the constant diminu- 

 tion of the volume of the water, and per- 

 fectly independent of the different weights 

 of the two vertical columns. The effect 

 would, indeed, be very similar to what 

 might be produced by a constant but 

 slight increase in the diameter of the 

 pipes, where the water would be continu- 

 ally moving into a larger and larger 

 space, and where of consequence it must 

 move much more freely than where the 

 space was constant in reference to the 

 bulk of the fluid." 



To circulate water to levels either 

 above or below the boiler, it is necessary, 

 in the first case, (fig. 349,) to have the 



Fig. 349. 



top of the boiler a fixed, from which a 

 pipe may be taken to any height and 

 made to descend again; but it must not 

 rise nor fall twice after leaving the boiler. 

 That part of the pipe b that is highest 

 above the boiler should have a small air- 

 tube inserted in it at the bend; for, 

 should air accumulate there, no circula- 

 tion can take place till the air is extracted. 

 In the next case, fig. 350 — namely, circu- 



Fig. 350. 



a 



lating water below the boiler (a) level — it 

 is necessary to raise the water as it is heat- 

 ed to as great a height (b) above the boiler 

 as it is intended to carry it below it (c). 



Much has been said on the necessity of 

 supply -cisterns for feeding the boilers, 



