68 yidmfnfurtment of the Intcnfiues of Light, 



will b« illuminated by the fmallcr light only, and reverfcly the other fliadow will be iriu- 

 minatcd by the greater liglit. That is to %, in (hort, the (Ironger light will be attended 

 with the deeper fliadow. But it is cafy, by removing the greater light to a greater dif- 

 tance, to render the illumination it produces at the commorj furface equal to that 

 afFordcd by the lefs. Experiments of this kind may be conveniently made by faften- 

 ing a flieet of white paper againft the wall of a room. The two lights or candles in- 

 tended to be compared, muft then be placed fo that the ray of light from each flial! fait 

 with nearly the fame angle of incidence upon the middle of the paper. In this fituation, 

 if a book or other objeft be held to intercept part of the light which would have fallen 

 on the paper, the two fliadows may be made to appear as in Figure lo, Plate V, where A 

 reprefents the furface illuminated by one of the lights only ; B, the furface illuminated by 

 the other light ; C, the perfect lluidow from which both lights are excluded. It will 

 cifily be undcrftood that tlie lights about D and K, near the angle F, will fall with equal 

 incidences when the double (hadow is made to occupy the middle of the paper ; and con- 

 fcquently, if one or both of the lights be removed dire£lly towards or from the paper, as 

 the appearances may require, until the two (hadows at E and D have the fame intenfity, 

 the quantities of light emitted by each will be as the fquaree of the diflances from the paper. 

 By fome experiments made in this way in the year 1785, I was fatistied that the degree of 

 illumination could be thus afcertained to the eightieth or ninetieth part of the whole. 



By experiments of this kind many ufeful particulars may be fliewn. Thus, for example, the 

 light of a candle, which is fo exceedingly brilliant when firll fnuffed, is very fpeedily diminilhed 

 to one-half, and is ufually not more than one-fifth or one-fixlh before the uneafinefs of the 

 eye induces ns to fnufF it. Whence it follows, that if candles could be made fo as not to 

 require fnuffing, the average quantity of light afforded by the fame quantity of combufti- 

 ble matter would be more than doubled. In the fame way, likewlfe, fince the coft and. 

 duration of candles, and the confumption of oil in lamps, are eafily afcertainable, it may be 

 (hewn whether more or lefs of light is obtained at the fame expence during a given time, by 

 burning a number of fmall candles inftead of one of greater thieknefs. From a few ex- 

 periments already made out of the numerous and ufeful feries that prefonts itfelf, I have 

 reafon to think that there is very much wafte in this expenfive article of accommodation. 



In the lamp there are three articles which demand our attention, the oil, the wick, and 

 the fupply of air. It is required that the oil fliould be readily inflammable, without con- 

 taining any fetid fubftanoe which may prove often five, or mucilage, or other matter, to 

 obftruil the channels of the wick. I do not know of any procefs for meliorating oils for 

 this purpofc, excepting that of wafliing with water containing acid or alkali. Either of 

 thcfe is faid to render the mucilage of animal oils more folubic in the water; but acid is 

 preferred, becaufc it is lefs difpofed to combine with the oil itfelf. The office of the wick 

 appears to be chiefly, if not folely, to convey the oi' by capillary attraftion to the place of 

 combuftion. As the oil is confumed and flies off, other oil fuccecds, and in this way a 

 continued current of oil and maintenance of the flame are effefted. But as the wicks of 

 lamps arc commonly formed of combuflible matter, it appears to be of fome confequence 

 what t!ie nature and ftrufture of this material may be. It is certain that the flame afforded 

 by a wick of rufli differs very confiderably from that afforded by cotton; though perhaps this 

 difference may, in a great mcafurc, depend on the relative dimcnGons of each. And if 



