a) 
320 
and with the reduction of the stress, the curve 
drops, until very shortly the specimen is rup- 
tured, and the apparatus comes to a standstill. 
The other three curves are those given by a 
piece of boiler-plate and of a specimen of 
muck-bar, and are very good examples of the 
‘value of the autographic method. As is well 
known, both boiler-plate and muck-bar are 
decidedly non-homogeneous; and as a result 
we have curves here that are exceedingly 
irregular, especially after passing the elastic 
limit. While they bear a general resemblance 
to the previous ones, they are full of points 
of inflection, turning and twisting about, and 
giving one an idea that the specimen consisted 
of a bundle of threads or fibres which gradually 
parted under the action of the stress, giving 
any thing but a constant and uniform action. 
In conclusion, a word as to the practical ac- 
curacy of the lever testing-machine may not be 
out of place. The machine under considera- 
tion has been subjected to severe use for nearly 
two years, during which time its sensitiveness, 
even when loaded, has not risen so high as the 
least reading on the poise. From this, and 
from long practice in similar scale-work, it may 
be safely stated that the testing-machine, with 
proper care, may have an exceedingly long life. 
The attainment of absolute accuracy in any 
department of investigation, would, if it were 
possible, be an extremely desirable result ; yet 
even our best experiments are simply close ap- 
proximations to the truth, and it will be granted 
that it is desirable to make all of our improve- 
ments commensurate towards an absolute stand- 
ard of accuracy. It is of no importance to 
carry the weighing-power of the testing-ma- 
chine beyond the possibility of the measure- 
ment of the bar. Forexample: supposing the 
tests most frequently made are those of bars 
-having about a square inch of cross-section. 
In a piece of iron an error of a thousandth of 
a square inch of cross-section corresponds to a 
possible inaccuracy, in the stress produced on 
the bar, of fifty pounds ; while the correspond- 
ing quantity in a steel bar corresponds to about 
seventy to ninety pounds. There are very few 
lathes in the country in which it is possible 
to turn a bar so exactly that it shall be per- 
fectly round, and that there shall be no varia- 
tion from one end to the other of more than a 
thousandth of a square inch. There are few 
men that are capable of manipulating any lathe 
to produce such a result; and there are still 
fewer gauges that are capable of measuring 
even a perfect bar so as to exclude the possi- 
bility of an error as great as a thousandth of 
a square inch. Now, if it be impossible to 
SCIENCE. 
(Vox. IIL, No. 58. 
measure our bars to within an error of fifty 
to a hundred pounds in the testing-machine, is 
it of any importance to refine the machine 
beyond this reading? In the lever system 
of testing-machines it is perfectly possible to 
obtain a machine which will uniformly and 
constantly give readings which shall not have 
a greater variation than from five to twenty 
pounds ; and, if our bars can only be measured to 
fifty or a hundred pounds, would it not be wiser 
to spend money in refining the gauges rather 
than in refining the machine? Again: when 
the consideration of the tests on the full-sized 
members occurs, or bars direct from the rolls, - 
carrying with them the scale, and other imper- 
fections from the mill, the possibility of meas- 
uring to a thousandth of a square inch becomes 
absurd, and two or three hundredths is the 
nearest approximation that can be made. In 
making a test of an ordinary I-bar, of, say, 
five or six inches of cross-section, it is certain 
that the bar has any thing but an absolutely 
uniform section from end to end ; and how long, 
may it be asked, would it take to measure that 
bar from end to end, so that the least cross- 
section could be obtained for the record? And 
again: in actual experience it has been fre- 
quently found, that, having obtained what is 
supposed to be the least cross-section, the test- 
piece may break in a totally different place. 
It will be conceded that practical engineers 
care very little for test-records beyond the hun- 
dredth’s place of figures ; and what the country 
wants at the present time, is not so much test- 
ing-machines constructed with a theoretical 
refinement of accuracy, as a large number of 
practical machines, so that one may be located 
in every iron-works in the country, and means 
to carry on the experiments and to obtain 
from these machines a practical knowledge of 
what America’s constructive materials really 
are. A. V. ABBOTT. 
NEW METHOD OF MOUNTING REFLEC- 
TORS 
Ir is well known to all who have given at- 
tention to this subject, that the optical perform- 
ance of great reflecting-telescopes has not 
been proportional to their size, and that the 
mechanical difficulties of keeping a large re- 
flector in proper figure in different positions 
have been apparently insurmountable. A plan 
of supporting a large mirror, devised by Mr. 
Henry, has been adopted in Paris, which it is 
hoped may obviate this difficulty. It consists, 
in principle, in supporting the mirror upon a 
1 Extracted from a report to the secretary of the navy on 
improvements in astronomical instruments. 
