February 24, 19 16] 



NATURE 



721 



ction passing through the centre of a rivet in a plate 

 re shown in Table II., for the case of a plate in 

 hich both the overlap and the widths of metal on 

 uh side of the rivet arc equal to the diameter of the 



\et. 



Stress r..tios 

 ria I '20 



{p,+Pr)l/> 2-55 

 ifit-pr)lp 2 -80 

 f,lp ... 268 

 /,// ... -0-125 



rfa ... 1-40 

 ipt+pr)/p -4-31 



(pt-PrVp - 2- S6 



p.lp ... -3-44 



t'Jf) 



Table 1 1 . 



Stresses across the sectic 

 1-40 1-6) 180 2-00 

 220 I -84 1-47 I 18 



I 60 I 00 076 os6 

 1-90 142 I 1 1 087 

 0-30 0-42 035 0-31 



Stresses on centre line below rivet 

 I '50 I 70 I 90 2 20 

 •2-32 -090 -0-37 0-4? 

 232 -I 87 -I 48 -I 25 

 232 -1-39 -0-93 -0-41 

 — 0-49 0-55 0-84 



/,.// ... -0-88 



In these determinations the distance r of the point 

 examined is measured from the centre of the rivet 



Fic. 4 — Load applied by rivet. Piiniip.nl stres.ses. 



in terms of its radius a, while the stress pr across the 

 section and the stress pt in the section are given in 

 terms of the mean stress. 



The experimental values of the principal stresses are 

 \en in Fig. 4. They show that the tensile stress 

 a the cross section reaches a high value, 

 while below the rivet an even greater com- 

 pression stress is produced. The measurements of 

 radial stress along the sections chosen give marked 

 compression close to the rivet, and it is worthy of note 

 that they are very nearly zero at the outer boundaries 

 of the plate, results which confirm the general accu- 

 racy of the measurements. Other measurements of a 

 similar kind show that the action of a rivet produces 

 an intense stress at the hole, sometimes reaching five 

 times the stress in a full plate. In a transparent 

 model this is often accompanied by {lermanent over- 

 stress and local yielding, which latter tends to equalise 

 Mie stress in the material. 



Blocks in Compression. 

 A problem of considerable importance in practical 

 engineering is the distribution of stress in a rect- 

 angular block subjected to pure compression. This 

 case occurs in the testing of materials when equal and 

 v)i)positely directed loads are applied to the parallel 

 faces of a short rectangular block, of a material like 

 stone, brick, or concrete, to give a compressive stres* 

 as uniformly applied as is possible. 



The manner in which the load is applied to the end 

 faces of such a rectangular block is known to exert 

 an influence upon the distribution of stress and strain, 

 and to obtain consistent results the end faces of small 

 specimens are sometimes faced by aid of a surface 

 plate in order to give a uniform bt'aring. In a large 

 .specimen, where this process becomes very laborious,, 

 it is often faced with plaster of Paris, and plates of 

 millboard are sometimes interposed in addition, to 

 give a uniform bearing area between the pressure 

 plates of a testing machine and the specimen. 



It is difficult to determine experimentally the distri- 

 i>ution of stress in a short block of an engineering 

 material, but if a transparent model is used the 

 problem is comparatively simple ; and it may there- 

 tore be of interest to indicate what ])rogress has been 

 made in the solution of this practical problem by 

 Prof. Filon and myself. 



In the majority of testing machines the mechanical 

 arrangements for ai)plying a pure compression load 

 are defective, and for' investigations of* this kind a 

 special form of testing-machine was constructed 

 to give a very accurately applied load ; and, 

 in order to facilitate the measurements, it was found 

 convenient to suspend the lateral extensometer, X, 

 from a three-point support (Fig. ^), borne bv a block 

 sliding upon the vertical pillars P,, Pj of the testing 

 machine, and having twO cross horizontal slides B and 

 C at right angles, these latter being capable of 

 adjustment by micrometer screws D and E, in 

 such a manner that the measuring instrument could 

 be moved to any position with respect to the speci- 

 ^ men B, and the horizontal and vertical co- 



ordinates detennined to an accuracy of 

 I / 1000 in. 



In the photograph now shown a block of 

 square section is subjected to compression stress, 

 and the optical effects show that the stress is far 

 from uniform, although all possible precautions were 

 taken to ensure a j)erfectly uniform bearing. It appears 

 to be of much less intensity at the end faces than 

 at the centre of the block. The measurements con- 

 firm this, and show that there is a very considerable 

 end effect, tending to reduce the compression stress 

 at the centres of the end faces, and only disappearing 

 at a moderate distance away. In work of this kind 

 where it is necessary to examine the effect of the 

 pressure of an opaque body upon a transparent onC; 

 there is considerable difficulty in accurately measur- 

 ing the stresses very close to the boundary between 

 the surfaces. 



The diagram now shown of some preliminary 

 measurements indicates the general character of the 

 distribution, in which the curves denote the vertical 

 tx^mpression stresses at various distances from the end 

 faces. 



The non-uniform character of the stress distribution 

 is evidently due to the way in which the pressure is 

 applied to the material under test, and in this example 

 the brass pressure plates prevent free lateral displace- 

 ment of the material under load. If, therefore, a 

 more extensible material is interpo.sed for tran.s- 

 mitting the load to the block, we may expect to 

 obtain greatest stress at the centre, and this is actually 

 what occurs. 



VO. 



2417, VOL. 96] 



