234 THE SCIENTIFIC PAPERS OF 



difcure of 4 horse-power. The experiments already referred to 

 show that the most effective height at which to place the naked 

 electric arc of 1,400 candle-power is about 2 metres. By using a 

 metallic reflector, the major portion of the upward rays may be 

 thrown down upon the surface to be illuminated, and that height 

 may be taken at 3 metres. If an electric arc of 6,000 candles 



was employed, the height would be ~/T~A (] O x 3 = 6*2 metres, at 



which such an electric light should be fixed. In operating upon 

 an extended surface, several lamps should be so placed as to make 

 the effect over it tolerably uniform. This would be so if the 

 radiating centres were placed at distances apart equal to double 

 their height above the ground ; for a square foot of surface mid- 

 way between them would receive from each centre one-half the 

 number of rays falling upon such a surface immediately below a 

 centre. A plant at the intermediate point would, however, have 

 the advantage of presenting a larger leaf surface to both sources 

 of light ; and to compensate for this advantage, the light centres 

 may be placed yet further apart, say at distances equal to 3 times 

 their elevation, or 18 metres. Nine lights so placed would suffice 

 for an area 54 metres square, or about f acre. If a high fruit 

 wall were to enclose this space, this will also get the full benefit of 

 electric radiation, and would serve at the same time to protect the 

 plants from winds. By subdividing the area under forced cultiva- 

 tion by vertical 'partitions of glass, as has been done with excel- 

 lent results by Sir William Armstrong, protection is insured 

 against injury from this latter cause. 



There would be required to maintain this radiant action a 

 9 x 4 = 36 horse-power engine, involving the consumption of 

 36x2| = 90 pounds of fuel per hour, which, for a night of 

 12 hours, with 40 pounds for getting up steam, amounts to half-a- 

 ton, costing, at 16s. a ton, 8s. a night. This does not include, 

 however, the cost of carbons, or of an attendant, which would 

 probably amount to as much more, making a total of 16s. If, 

 however, an engine could be utilised doing other descriptions of 

 work during the day, the cost of steam power and attendance for 

 the night work only would be considerably reduced. 



I have assumed in the calculation just given the use of fuel to 

 produce mechanical energy, but the question will assume a totally 



