104 NATURAL HISTORY. [CH. V. 



must appear evident how necessary it is to get rid 

 of it as soon as it disengages itself, and continually 

 and gently to renovate the atmosphere. 



Light. — Many think that light is injurious to silk- 

 worms. It is certain that in their native climate it 

 does not injure them, although they are exposed to 

 it by various circumstances ; however, there is here 

 no question of exposing them to the sun, but only 

 of rendering their habitations as light as our own. 



It is always observed, that on the sides on which 

 the light shines directly on the hurdles, the silk- 

 worms are more numerous and stronger than in 

 those places where the edges of the wicker hurdles 

 intercept the light, and form a shade, Avhich is also 

 a reason for having very low edges to the wicker 

 trays. Even the sun shining full on the worms 

 seems not to annoy them. If the rays are too hot, 

 and shine too long on them, they may suffer; but 

 this cannot occur, nor does it affect the question, 

 as it is not proposed to expose the silkworms to the 

 sun, but only desired to show that the air is more 

 vitiated, and that there is more damp in a dark 

 laboratory than in a light one. 



The effect of this, perhaps the most powerful 

 agent on life in general, should be particularly 

 attended to in the rearing of the silkworm. 



In order, therefore, to cultivate the silkworm with 

 advantage, and to put this precious insect in nearly 

 the same circumstances as if it were in its native 

 climate, it is necessary that ventilators should be so 

 placed that the mass of noxious vapours should 

 never be allowed to stagnate to the prejudice of the 

 silkworm ; that stoves should be so situated as to 

 maintain an equable and a fit temperature, and that 

 the windows should be sufficient to admit a due de- 

 gree of light. 



The size of the laboratory will vary according to 

 the number of worms to be reared ; but in all cases 

 the space should be ample, as nothing is more pre- 



