( It ) 
mofs at one end of the box, fo as not to be fhook out of their 
places, and laths fhould be nailed acrofs the inlide to fupport their 
branches, and keep them from preffing upon one another. The 
beft mofs that can be ufed on thefe occafions is the Sphagnum pa- 
lujire , or fwamp mofs, which is very foft, whitilh, and fpongy; 
it will retain water a long time, and not be liable to putrefy. 
The following method of preferving feeds from turning rancid 
from their long confinement, and the great heat of the climates 
which they mull; necelfarily pafs through from China, was com- 
municated to me fome years ago by the celebrated Profeflbr 
Linnaeus, of IJpfal, in Sweden. He advifes, that each fort of 
feed fhould be put up in feparate papers, with fine fand among 
them, to abforb any moillure (dried, loamy, or foapy earth may 
be tried): thefe papers, he fays, fhould be packed clofe in cylindri- 
cal glafs, or earthen veffels, and the mouths covered over with a 
bladder, or leather tied faff round the rims : he then directs that 
thefe yeffels, with the feeds in them, fhould be put into other 
veffels, which fhould be fo large, that the inner veflel may be co- 
vered on all fides, for the fpace of two inches, with the following 
mixture of falts. Half common culinary fait; the other half to 
confift of two parts of falt-petre, and one part of fal-ammoniac, 
both reduced to a powder, and all thoroughly mixed together, to 
he placed about the inner veflel, rather moift than dry. This he 
calls a refrigeratory; and fays it will keep the feeds cool, and hin- 
der putrefaction. Perhaps if fmall tight boxes, or calks or bottles 
of feeds were inclofed in calks full of falts, it might be of the fame 
ufe, provided the falts do not get at the feeds ; and as fal-ammo- 
niac may not be ealily met with, half common fait, and the other 
half falt-petre, or common fait alone, might anfwer the fame end. 
But it would be very necelfary to try both methods, to know whe- 
ther the latter would anfwer the purpofe of the former, as it would 
be attended with much lefs trouble, and might prove a ufeful 
C 3 method 
