when driven into different species of Timber. 41 
oak, and equally seasoned locust. ‘Thus the weight which in those 
three cases drew the square spike from chestnut, was 1995; and 
that which extracted the broad flat one 2394; and that which 
drew the narrow flat one from the same timber was 2223. The 
sum of these is 6612. The sum of the three numbers for the same 
three spikes used with oak, was by experiments 5, 6, and 7, 13110; 
and the sum of the three in locust, by experiments 13, 14, and 15, 
is 18280; these three numbers have to each other the relation of 
1, 2, and 22; from which we infer that oak is almost precisely twice, 
and locust 23 times as retentive as unseasoned chestnut. By com- 
paring together the results of experiments | and 2, it will be seen 
that the weights required for extracting the two spikes respectively, 
are more nearly proportional to the breadth than to either the thick- 
nesses, or the weights of the spikes. For the spike with a breadth 
of .405 inch and a thickness of .402, required 1995 Ibs. for its 
removal, while that which had a breadth of .375 inch took 1873 
lbs. Now .373 : 405::1073 : 2033 for the calculated retentive- 
ness, instead of 1995, as given by experiments ;—a difference of 
only +388 lbs. between the observed and calculated results. Cal- 
culating the retention by the weights of the respective spikes, we 
should have 866 : 942::1873 : 2987, ora difference of 42 lbs. 
while using the thickness alone, we obtain .884 : 402::1873 : 1960, 
a difference of an opposite kind of 35 Ibs. from the observed result, 
the greater thickness yielding the less retentive power. ‘This cor- 
respondence between the breadths and the extracting weights be- 
comes still more apparent when we compare the third, and especially 
the fourth with the second experiment, Thus ie the broad flat 
spike, (3d Ex. et a = experiment 2, we obtain 
By breadths, — .373:539: : 2701, instead of 2304, pers 
weights, - .866:898: ipo: 3; 1942, 
© thicknesses, .384 : 288 : :-1873 : 1379 as “ 
and for the thinner and lighter spike, (Ex. 4th. aaa with 
the same, 
By breadths,  .373:390::1873: 1958, nee aaa, observed diff. —-265 
“ weights, .866:566:: 1873: 122 Mies =F «2909 
“ thicknesses, .384; 253 :: 1873: 1234, os * = «989 
Nearly the same conclusions would result from a comparison of 
those trials, which were made on seasoned white oak and locust. 
Indeed, it appears that with a given breadth on the face of the spike, 
a diminution of thickness is sometimes a positive advantage to the 
en ee of the timber ; for in white oak, the — which had 
6 
XXII.—No. 1 
