from the Cold induced by tJie Evaporation of Water. 101 



interval of about 90° on Fahrenheit's scale. Putting, therefore, 

 PH = / — ^' = D, we get 



[^'-_64 + >n/(^-64)«+2025]' — I> * 



So far as the Table has been compared with observation, 

 this expression keeps very close to it ; and even in its extreme 

 case, where ^ =r 110°, and D = 34°, this formula makes the 

 dew-point 66°. 9, which is just 1° higher than in the Table. 

 At low temperatures, the difference is greater, though only on 

 greater values of EI ; but, in such cases, it would require expe^ 

 riments to decide how far any of them may be right *. But, 

 independently of its use for comparing the two methods, this 

 formula, since it does not require the aid of any Tables of the 

 force, &c. of steam, will, (where it is wished to use a formula), 

 be found a convenient one for directly computing the dew- 

 point ; and will be still more easy, if, in place of the difference 

 of the squares in the denominator, we use the equivalent pro- 

 duct of the sum and difference of the quantities without squar- 

 ing. A similar formula, as also a similar diagram, would give 

 the temperature at which vapour of the same density as that in 

 the air would be at its maximum density, or in a state of satu- 

 ration ; but a method of obtaining both that and the dew-point 

 from one diagram or plate will be noticed presently. 



Several of the lines in the diagram being only for illustrating 

 the foregoing investigation, would fall to be omitted altogether 

 on a large scale. But to render the project more complete, the 

 space between the two hyperbolas should be occupied by lines 

 parallel to AG ; and at the distance of a degree one from another, 

 or each marking a degree of temperature. Close to the right 

 side of the hyperbola EO, a set of divisions might be put, to 



• Although it were turning out, from more extensive observations, that 

 the hyperbolas would not exactly answer for all temperatures, yet, from their 

 answering so well, as we have just seen they do, throughout the range compared 

 with observation, there seems little reason to fear, that no curves whatever 

 would suit, or even that the proper curves would differ much from hyperbo- 

 las. The effect could no doubt be considerably modified, by using two hy- 

 perbolas not exactly equal ; as also, by shifting a little the position of the 

 straight line CD, and of the parallel marked 54°, &c., and perhaps by slightly 

 curving these lines. But it will be soon enough to have recourse to any such 

 ex])edtent8 when they may be seen to be necessary. 



