152 History of Mechanical Inventions and 



supply of itself, does away all this ; and the quantity, once agreed upon, 

 will continue regular and uniform. 



AB, is a basin of water immediately behind the tunnel of the reservoir, 

 in which basin the water is always kept at the same level, by the appara- 

 tus formerly described. 



BC, one of a number of sluices of the same kind on that basin that 

 turns upon pivots at C. 



DE, a can, open at top, and having a very small aperture in its bottom. 

 F, a pulley. 



DFG, a chain, which, passing over pulley F, has one end fixed to sluice 

 BC, and the other to can DE. 



H, a weight that keeps the sluice BC always shut, when the can DE is 

 empty ; when that can is full of water, it lifts weight H, and opens the 

 sluice. 



IK, a section of the rivulet, immediately before it falls into the reser- 

 voir. 



s LMN, a pipe which communicates between the rivulet at IK, and the 

 can DE. 



Z, this letter cannot be shown on the drawing, but is the supposed 

 mill or factory, on the rivulet below the reservoir, that requires the great- 

 est quantity of water. 



Now, suppose the weather to be dry, and the water so low in the rivu- 

 let as only to reach the aperture 1, then the water that flows out there 

 passes down pipe L MN, and runs out at N into can DE, which, being 

 thus filled with water, opens sluice BC, which passes as much water as 

 the rivulet then brings into the reservoir.* But, when the rivulet swells 

 so as to flow out at aperture 2, then (the opening at N not being able to pass 

 the whole) the water rises in pipe LM, and passes along pipe OF, and, 

 falling into another can, opens a second sluice, which, with the first, pass- 

 es as much water as the rivulet then brings into the reservoir. When 

 the w ater in the rivulet rises so as to flow out at aperture 3, it rises also 

 in LM, and, passing along pipe QR, flows out at R into a third can, and 

 opens a third sluice, and these three pass as much water as the rivulet 

 then brings, and which is here supposed to be the greatest quantity want- 

 ed at the place Z. Suppose, now, the flood should still continue to in- 

 crease, the streams and surface water between the reservoir and Z will 

 increase the rivulet at Z, as well as the higher streams increase it at IK ; 

 but there was formerly enough of water at Z : when, therefore, the rivu- 

 let rises so as to flow out at aperture 4, the water will rise also in the tubes 

 NS, PT, RU, till it come to float S, which it lifts, and thereby shuts 

 valve N ; the water in can DE then passes out at the small opening in its 

 bottom, and the weight H shuts sluice BC, which stops as much water in 

 the reservoir as the streams below have increased. When the water rises 

 in the rivulet, so as to flow at aperture 5, it rises also in the tubes till it 



* By a very simple contrivance any one can tell, by merely looking at them, 

 whether the same quantity of water is flowing from and into the reservoir. 



