NA TURE 



[May 5, i88 



abrading purposes. It is known however that if a piece of this 

 steel be heated, and then suddenly cooled (generally in a bath of 

 water or oil) it becomes much harder, not only on the surface, 

 but throughout, provided its thicl<ness be not exc-ssive. The 

 greater the range of thi^ cooling (in other words, the difference 

 between the temperatures of the steel and the bath at the first 

 moment of "quenching") the more intense is the hardening, but 

 at the same time the greater the brittlene^s of the piece. Hence it 

 is always desirable that the range of cooling should be as small as 

 is consistent with the steel acquiring that degree of liardness which 

 is essential for the w )rk it has to do. This condition is secured 

 by the further operation of tempering. In this process the steel 

 is first hardened to excess by rapid coaling, then re heated with 

 o-reat care to a certain temperature corresponding to the purp jse 

 for which it is intended, and then quenclied again from that 

 temperature. The particular point at which to stop the re-heating 

 is recognised by one particular hue in what are called "the 

 colours of tempering," i.e. a fixed range of colours, commencing 

 with pale yellow and ending with darlc blue, which the steel is 

 always seen to assume in succession as its temperature gradually 

 rises. Thus, if the article in question be a sword it is heated to 

 a brio'ht blue ; if it be a cold chisel it is stopped at a brownish 

 orange. 



The various attempts to explain these singular facts (at least on 

 the part of French and English metallurgists) are set forth in the 

 Committee's Report. In the first place it seems now to be gene- 

 rally held that pure steel is a compound of iron and carbon only, 

 and that these two elements exist, not in a state of chemical 

 combination (formmg some definite carburet of iron), but of 

 intimate mechanical mixture, such as chemists call by the 

 name of "solution." The question next arises, What is the 

 exact condition of each of these independent elements. In the 

 case of very soft, or "grey," cast-iron, it is known that the 

 carbon is not wholly in solution, but occurs partly in molecules 

 of pure graphite. Following this hint M. JuUien has advanced 

 the theory that molten cast-iron, or molten steel, is a solution of 

 liquid carbon in liquid iron ; that under slow cooling part of the 

 carbon separates as graphite, while the remainder continues in 

 solution ; but that with rapid cooling this separation does not 

 take place, and the whole of the carbon crystallises, forming, 

 when cool, a "sdution" of crystallised carbon in amorphous 

 iron. This view of the difference between hard and soft cast- 

 iron, or hard and soft steel, is accepted by Caron, Akerman, 

 and others, partly on the ground that hardened steel dissolves 

 completely in hydrochloric acid, while the same steel, after 

 annealing, will leave a residue of insoluble carbon. Jullien, how- 

 ever, goes beyond this, and would explain the whole phenomena of 

 hardening on the same principle. He holds that carbon liquefies 

 in presence of red-hot iron, and is absorbed by it ; that if the 

 mixture is cooled slowly the carbin remains amorphous, but if 

 cooled quickly the carbon crystallises in the diamond form ; and 

 thus hard steel is iron set m a matrix of diamond. This theory, 

 though ingeniously supported, labours under the difficulty that 

 the liquefaction of carbon has never been otherwise achieved; 

 and also that it gives no explanation whatever of the phenomena 

 of tempering, especially the characteristic colours. On the 

 other hand, Barha and Akerman hold that the hardening of steel 

 is due to the severe com^jression produced in the outer layers by 

 the contraction in rapid cooling ; this compression at once retain- 

 ing a greater proportion of carbon in solution, and rendering the 

 whole mass more physically dense and compact. But the Report 

 points out that the outside layers, which are the hardest, are 

 brought into a state of tension, not compression, owing to their 

 inability to contract over the hotter mass inside. 



For these reasons the Committee have rejected both these 

 theories, and propose one of their own, due apparently to Mr. 

 William Anderson, but suggested by the experiments of Edison 

 on platinum wire, an account of which appeared in Nature, 

 vol. XX. p. 545. Thev ref -r *o the generally-accepted fact that 

 ordinary steel contains a certain propor ion of occluded gases 

 (consisting, according to Miiller, of hydrogen, nitrogen, and 

 carbonic oxide). They suggest that the application of heat 

 causes these "ases to l)e expelled through minute fissures which 

 open in the steel, as they opened, according to Edison, in the 

 platinum wire observed liy him ; and that sudden cooling prevents 

 the re-absovption of what has been expelled, perhaps actually 

 tends to exjwl the remainder. By the loss of the-e gases the 

 metal becomes dei ser ai,d harder than before. If the metal be 

 now expanded oy ende hrating, the fissures open, re-absorption 

 begins ; and the various changes which the surface undergoes 



during this process are marked by the succession of colours 

 which are characteristic of tempering.' The Committee propose 

 to make a .series of experiments to test the truth of this theory, 

 which is certainly ingeni lus, and if confirmed would go far to 

 remove the difficulties which beset this important subject. 



/^orm of Rivetedyoints. — The second Report is written by Prof. 

 W. C. Unwin, of Cooper's Hill, and is on "the best form of 

 riveted joints to resist strain in iron or steel, or in combination." 

 It may be well to explain that by a "riveted joint " is meant the 

 mode of fastening together the strips, or plates, out of which 

 boilers, tanks, girders, and other structures in wrought iron or 

 steel are built up. This mode in its simplest form is as follows : — 

 A row of holes is punched or drilled along the edge of the two 

 plates to be united, and these edges are then made to lap over 

 each other so that each hole in one plate comes fair with a hole 

 in the other. A red-hot rivet (that is a pin with a rounded head) 

 is then passed through both holes, and the end is flattened down 

 by hammering or pressure, so as to form a second head. The 

 two plates are thus pinned together by the rivets, and so long as 

 these remain entire they cannot be separated without tearing 

 across. Such a joint is called a "single-riveted lap-joint." 

 It is obvious that the plate must be greatly weakened by the 

 piercing of the holes ; and as a matter of fact it appears that 

 such joints cannot be arranged to give more than about half the 

 strength of the solid plate. To increase this "proportion of 

 strength," as it is termed, the rivets are sometimes arranged in 

 two or three rows, and of course more widely spaced in each 

 row ; or the edges of the two plates are simply brought up 

 against each other and secured by either one or two "cover- 

 strips " fitting over and riveted to both. 



Each of these forms demands a separate investigation in 

 order to fix its design. Thus a single-riveted lap-joint under a 

 tensile stress may fail in any one of the following ways : (i) 

 the rivet may cut into the plate, enlarging and injuring the hole; 

 (2) the plate may cut into the rivet, and finally shear it off; (3) 

 the part of the plate between the rivet and the edge may break 

 through, allowing the rivet to come away from the plate ; (4) 

 the plate may simply tear across along the line of rivets. It is clear 

 that in a perfect joint the dimensions must be such that the 

 resistance to each of these modes of fiacture should be the same, 

 and should have its greatest possible value. This could be easily 

 arranged if the absolute resistance of the material to these 

 various forms of stress were accurately known. But the values 

 of these resistances are of course very different for steel and for 

 iron ; they also vary considerably, whether in steel or iron, 

 according to tiie quality, and to some extent according to the 

 thickness. Hence experiments on these values become abso- 

 lutely necessary before any correct design can lie made out. 



The course which such experiments should take is fairly 

 sketched out by the author of the Report now before us. A 

 good and uniform quality of iron or steel, as the case may 

 be, should first be selected, both for plates and rivets. The 

 resistance of this material to the various forms of stress should 

 then be carefully ascertained by experiments made both with 

 simple bars or plates, and also with actual riveted joints, so 

 designed that they shall be certain to fail in one particular way. 

 These constants once settled, it is easy to calculate for any 

 description of joint the dimensions which will give the highest 

 proportion of strength. A joint should then, and not till then, 

 be prepared, having exactly these dimensions, and a few others 

 having dimensions varying slightly from these in each direction. 

 If, on testing, the first joint proves to gives the highest breaking 

 strain of the set, thexorrectness of the whole investigation \vill 

 be established. 



Unfortunately the method thus sketched out has not hitherto 

 been adopted. The immense practical importance of the sub- 

 ject (for the money expended yearly on riveted structures may 

 be counted by millions) has indeed brought forward a host of 

 experimenters ; and the mere classification and abstracting of 

 their re-ults occupies no less than sixty octavo pages of this 

 Report. But almost without exception they seem to have begun 

 at the wrong end, i.e. they started with making a riveted joint 

 of what they chose to consider to be the best design, and then 

 pulled it asunder. In addition, scarcely any of their experiments 

 have been made with the care and accuracy, or on the scale, 

 which the subject demanded. In fact it is not going too far to 

 say that 90 per cent, of these experiments are only injurious, as 



' Thj Report supposes that the colours of tempering are due to diffraction, 

 not to interference; this does not seem to be in accordance with the facts, 

 but it also does not' seem to be absolutely required by the exphnation. 



