490 
NATURE 
[Sepé. 17, 1885 
permanent injury results from such high stresses, because the 
number of repetitions is necessarily very limited. 
It appears natural enough to every one that a piece even of the 
toughest wire should be quickly broken if bent backward and 
forward to a sharp angle; but, perhaps, only to locomotive and 
marine engineers does it appear equally natural that the same 
result would follow in time if the bending were so small as to be 
quite imperceptible to the eye. A locomotive crank axle bends 
but 1-34th of an inch, and astraight driving axle the still smaller 
amount of 1-64th of an inch under the heaviest bending stresses 
to which they are subject, and yet their life is limited. During 
the year 1883 one iron axle in fifty broke in running, and one in 
fifteen was renewed in consequence of defects. ‘Taking iron 
and steel axles together, the number then in use on the railways 
of the United Kingdcm was 14,848, and of these, 911 required 
renewal during the year. Similarly, during the past three years 
no less than 228 ocean steameis were disabled ty broken shafts, 
the average safe life of which is said to be about three or four 
years. In other woids, experience has proved that a very 
mederate stress alternatirg from tension to ccmpression, if 
repeated about one hundred million times, will cause fracture as 
surely as a sharp} ending to an angle repeated perhaps only ten 
times. 
I have myself made many experiments with a view to elucidate 
the laws affecting the strength of iron- and steel-work subject to 
frequent alte:nations of stress. Perhaps the most suggestive 
series was one in which I subjected flat steel tars about 3 feet 
long, in pairs, to repeated bendirgs until one bar broke, and 
then testing the suviving bar urder Cirect tensile and cempress- 
ive stresses to ascertain to what extent the metal had deterior- 
ated. It had come under my notice, as a practical engineer, 
that if the compression members of a structure were unduly 
weak the fact became quickly evident, perhaps under the test 
load ; but if, on the other hand, the tension members were weak, 
no evidence might appear of the fact until frequent repetition of 
stresses during several years had caused them to fracture with- 
out any measurable elongation of the metal. In the case of 
crank-shafts, also, the fracture is invariably due to a tearing 
and not a crushing action. It appeared to me, therefore, 
eminently probable that repetition of stresses might be far more 
prejudicial to tension than to compression members, and, if so, 
the fact ought to be taken account of in proportioning a 
structure. 
This proved to be the case in my experiments. For example, 
the companion bars to those which had broken with 18,009 
reversals of a stress less than half the original breaking weight 
behaved, when tested as columns thirty diameters in length, 
precisely the same as similar bars which had done no work at 
all, whereas when tested in tension the elongation was reduced 
from the original 25 per cent. to 2°5 per cent., and the fracture 
appeared to indicate that the bars had been made of three 
different kinds of steel imperfectly welded together. With a 
stress reduced by one-fourth the number of bendings required to 
break the bars was increased to 1,200,000. In this instance the 
calculated maximum working stress on the extreme fibres was 
43 per cent. of the direct ultimate tensile resistance of the steel, 
and about 30 per cent. of the stress the tar was capable of sus- 
taining as a beam under the single application of a load. Of 
course, the bars failed by tension, and the extreme fibres had 
thus deteriorated as regards tensile stresses to the extent indi- 
cated by the above percentages. Tested as a column, however, 
the injury the bar had received from the 1,200,coo bendings was 
inappreciable. The ductility was of course very largely reduced, 
but ductility is a quality of comparatively little importance when 
a material is in compression. There is no ductility in the slender 
Gothic stone columns of our cathedrals, which, though heavily 
stressed, have carried their loads for centuries. As I found 
repeated bendings raised the limit of elasticity, I rather antici- 
pated finding an increased resistance from this cause in long 
columns. ‘This did not prove to be the case, nor did I find any 
difference in short columns four diameters in length. 
In addition to the preceding experiments with rectangular 
bars, I have tested the endurance of many revolving shafts of 
cast iron, wrought iron, and steel, with similar results. About 
5000 reversals of a stress equal to one-half the static breaking 
weight sufficed generally to cause the snapping of a shaft of any 
of the above materials. When the stress was reduced and the 
number of applications increased, I found the relative endurance 
of solid beams to be more nearly proportional to the tensile 
strength of the metal than to the breaking weight of the beam, 
a distinction of great importance where axles, springs, and 
similar things are concerned. Many of my experiments were 
singularly suggestive. Thus, it was instructive to see a bar of 
cast iron loaded with a weight which, according to Fairbairn’s 
experiments, it should have carried for a long series of years, 
broken in two minutes when set gently rotating. Also to find 
a bar of the finest mild steel so changed in constitution by some 
months of rotation as to offer no advantages either in strength 
or toughness over a new cast-iron bar of the same section. 
Although, as already stated, many more experiments are 
required before universally acceptable rules can he laid down, I 
have thoroughly convinced myself that, where stresses of varying 
intensity occur, tension and compression members should be 
treated on an entirely different basis, If, in the case of a tension 
member, the sectional area be increased 50 per cent. because 
the stress, instead of being constant, ranges from mz/ to the 
maximum, then I think 20 per cent. increase would be a liberal 
allowance in the case of a compression member. I have also 
satisfied myself that if a metallic railway bridge is to be built at 
a minimum first cost, and be free from all future charges for 
structural maintenance, it is essential to vary the working stress 
upon the metal within very wide limits, regard being had not 
merely to the effect of intermittent stresses, but also to the 
relative limits of elasticity in tension and compression members 
even under a steady load. 
Why an criginally strong and ductile metal should become 
weak and brittle under the frequent repetition of a moderate 
stress has not yet been explained. Lord Bacon touched upon 
the subject two or three centuries ago, but you may consider his 
explanation not wholly satisfactory. He said, ‘‘ Of bodies, some 
are fragile, and some are tough and not fragile. Of fragility, the 
cause is an impotency to be extended, and the cause of this 
inaptness isthe small quantity of spirits.” I am sorry to haveno 
better explanation to offer, but whatever may be the immediate 
cause of fragility, no doubt exists that it is induced in metals by 
frequent bendings, such as a railway bridge undergoes. This 
fact, however, is not recognised in our Board "of Trade Regula- 
tions, which remain as they were in the dark ages, as do those 
of the Ministry of Public Works of France and other countries. 
With us it is simply provided that the stress on an iron bridge 
must not exceed 5 tons per square inch on the effective section 
of the metal. In Fiance it is still worse, as the limiting stress 
of rather under 4 tons per square inch is estimated upon the 
gross section, regardless of the extent to which the plates may 
be perforated by rivet holes, In neither case is any regard had 
in the rules to intermittent stresses or the flexure of compression 
members. In Austria the regulations make a small provision 
for these elements ; and American specifications make a large one, 
the limiting stresses, instead of being constant at 5 tons, as with 
us, ranging from about 24 tons to 6} tons per square inch, 
according to circumstances. It is hardly necessary that I shouid 
say more to justify my statement that, as regards the admissible 
intensity of stress on metallic bridges, absolute chaos prevails. 
Engineers must remember that if satisfactory rules are to be 
framed, they, and not Governmental departments, must take the 
initiative. In former days the British Association did much to 
direct the attention of engineers to this important matter, but, so 
far as I know, the subject has been dropped for the past twenty 
years, and I have ventured, therefore, to bring it before you 
again in some detail. We are here avowedly for the advance- 
ment of science, and I have not been deterred by the dryness of 
the subject from soliciting your attention to a branch of science 
which is sadly in need of advancement. 
Had I been addressing a less scientific audience I might have 
been tempted rather to boast of the achievements of engineers 
than to point out their shortcomings. The progress in many 
branches of mechanical science during the past fifty years has 
exceeded the anticipation of the most far-seeing. Fifty years 
ago the chairman of the Stockton and Darlington Railway, when 
asked by a Parliamentary committee if he thought any further 
improvements would be possible on railways, replied that he 
understood in future all new railways would have a high earth- 
work bank on each side to prevent engines toppling over the 
embankments, and to arrest hot ashes, which continually set fire 
to neighbouring stacks, but in other respects he appeared to 
think perfection was attained. Shortly before the introduction 
of locomotives it was also thought perfection was attained when 
low trucks were attached to the trains to carry the horses over 
the portions of the line where descending grades prevailed, and 
all the newspapers announced, with a great flourish of trumpets, 
