228 REPORT 4, UNITED STATES ENTOMOLOGICAL COMMISSION. 
hopper less than it does to flour while its periphery is undermined, the 
outside of the column of poison will settle down best to where the blast 
can feed itself therefrom. In case dry poisons or mixtures that are not 
adhesive, and that jar through the sm^ll outlet of an ordinary hopper too 
easily are to be used, their flow may be regulated by a small plate adjust- 
able beneath the outlet and essentially on the plan described above. 
In such cases it is sometimes preferable to mount the plate on a single 
screw or support opposite and working through the blast in opposition 
to the small outlet of the hopper. 
Two other plans for hopper-supply are shown in Plate XXVIII, Figs, 
4 and 5. These also do not allow the powder to gravitate directly down- 
ward into the blast,/; but the hopper, P, is closed tight above by a re- 
movable screw-cap and narrows downwardly. The powder from its 
lower part has to be acted upon laterally by the blast, at the point o, 
in Fig. 4, and downwardly at o, in Fig. 5, where the hopper-throat curves 
around to open beneath. In either of these cases the communication, 
o, of the blast with the powder may be much elongated and can be ad 
justably opened or closed by a slide plate. The latter or the wall instead 
may have an area of many perforations, similarly located, or a slot or 
sflots for such powders as feed too strongly by other methods. These 
two forms of hopper I do not find so satisfactory for flour mixtures as are 
the forms with adjustable base slots previously noticed for feeding 
blasts. 
Finally, it should be added that powder can be fed by introducing 
in the vent of the hopper a moving reel, or brush, or rod, as has been 
shown in other machines, but these add complications and work less 
satisfactorily than the arrangement presented with the first feeder de- 
scribed above. 
The blower drum or casing is preferably made light, of stout zinc, 
brass or other sheet-metal. It need not be over a foot in diameter, and 
had better swell gradually in the direction of rotation to its ex-current 
orifice, as in Plate XXVIII, Fig. 1, <Z, although the true cylindrical 
form will answer. The head of the drum on one or on both sfdes should 
have a large central suction-opening equal to one-third or one half 
of its diameter and with.a stiff wire rolled in its margin. For making 
a blast through only one or tw© pipes the drum should be very short 
with the heads only 3 or 4 inches apart. This distance can be multi- 
plied by any number of pipes desired. 
Pipes for conducting the air blasts should be round, smooth, and 
bright. Tin suits well when not rusty, but is easily bent, unless of 
heavy grade. Zinc is preferable, if not acted upon by corrosive chem- 
icals, as it does not rust from the effects of atmosphere and moisture. 
.Oilcloth spouts kept expanded by a spiral of wire 18 feet long proved 
flexible, strong, and light, but too rough. Friction destroys the blast 
of the ordinary rotary blower very quickly. Crooks should also be 
avoided, and are much more objectionable in pneumatic blasts than in 
