Experimental Researches on the Strength of Pillars of Cast Iron , and other Materials .
By EATON HODGKINSON , Esq. Conmunicated by PETER BARLOW , Esq. , F.R.S. , 8sc .
Received April 22 , -Read May 14 , 1840 .
WVHEN we consider to what extent pillars of iron and of timber are used for the support of buildings , and reflect that there are no satisfactory rules by which to measure the strength of pillars , it becomes a matter of great importance to obtain such rules by means of experiment , and , if possible , to discover the laws on which they are founded .
A feeling of this kind , heightened by the remarks of Dr. ROBISON , in his Mechanical Philosophy , and by the strongly-expressed opinion as to our want of such knowledge by Mr. BARLOW* , led me to wish to undertake the inquiry .
I mentioned the matter , therefore , to my friend Mr. FAIRBAIRN , who , with that liberality which I have experienced from him on former occasions , at once put every means of a full investigation into my hands .
He expressed a wish that I should extend the inquiry to pillars of various kinds , ancient as well as modernn ; and leave no part of the subject in uncertainty for want of experiments sufficiently varied and extensive .
Thus freed from restraint , I endeavoured , in my wish to acquire the requisite information , to forget the expense to which I put my friend , and have made every effort to render the experiments at least ample , correct , and useful .
The pillars on which the experiments were made , were mnostly of cast irlon , as being the material in most general use ; but some were of wrought iron and steel , and others of wood .
In the earlier experiments , the pillars used were uniform cylinders , either with their ends rounded , so that the crushing force might pass through the axis , or with flat and parallel ends , so that the pillar , when placed between two perfectly parallel crushing surfaces , might have its ends completely bedded against them .
2 .
The pillars during the experiments were placed vertically , resting upon a flat , smooth plate of hardened steel , laid upon a cast iron shelf , made very strong , and lying horizontal .
The pressure was communicated to the upper end of the pillar by means of a strong lever acting upon a bolt of hardened steel , two and a half inches diameter , and about a foot long , kept vertical by being made to pass through a hole bored in a deep mass of cast iron ; the hole being so turned as just to let the bolt slide easily through without lateral play .
The top of the bolt was hemispherical , that the pressure from the lever might act through its axis ; and the bottom was turned flat to rest upon the pillar .
The bottom of this bolt , and the shelf on which the pillar stood , were necessarily kept parallel to each other ; for the mass through which the bolt passed , and that on which the shelf rested , were parts of the same large case of iron , cast in one piece , and so formed as to admit shelves at various heights , for breaking pillars of different lengths .
The case had three of its four sides closed ; circular apertures were , however , made through them , that the experimenter might observe the pillar without danger .
Fig. 1 , Plate XIII .
, will show more clearly the nature of the apparatus , and the manner in which the experiments were made .
Thus AB is the lever , about fourteen feet long , generally acting withi a leverage of from four to seven ; CD is an iron frame-work , in which the fixed end of the lever is placed , and in which it turns as on a pivot , the frame CD being fixed firmly in a strong heavy wall E F , part of which is imagined in the drawing to have been removed , to show the fixings and apparatus ; G is the steel bolt on which the lever acts ; H the socket through which it passes ; HIK the iron case admitting shelves as P , at different heights , for pillars of different lengths not exceeding five feet ; LM an additional part to be attached to the box HI K , for pillars of greater length ; N0 is one of the pillars resting upon the shelf P , and pressed upon by the pin G ; Q is the weight , which could be removed to any part of the lever , and increased or diminished at pleasure ; or entirely taken off the pillar , at any time , by means of the winding apparatus .
Experiments .
3 .
In order to ascertain the laws connecting the strength of cast iron pillars with their dimensions , they were broken of various lengths , from five feet to one inch ; and the diameters varied from half an inch to two inches , in solid pillars ; and in the hollow ones , the length was increased to seven feet six inches , and the diameter to three inches and a half .
My first object was to supply the deficiencies of EULER'S theory of the strength of pillars* , if it should appear capable of being rendered practically useful ; and , if not , to endeavour to adapt the experiments so as to lead to useful results .
4 .
As the results of the experiments were intended to be compared together , it was desirable that all the pillars of cast iron should be from one species of metal ; and the description chosen was a Yorkshire iron , the Low Moor , No. 3 .
The pillars were mostly made cylindrical , as that seemed a more convenient form in experiments of this kind than the square ; for square pillars generally break anglewise .
The experiments , in the first table annexed , were made on solid uniform pillars , rounded at the ends , that the force might pass along the axis ; and the metal was cast in dry sand , to obtain , as far as possible , uniformity in its texture .
In the second table , the pillars were uniform and cylindrical , as before , but had their ends flat and at right angles to the axis .
The pillars were from the same models as before , but were cast in green ( moist ) sand , which rendered the operation of casting less troublesome ; and it was conceived that the mode of casting would rmake little or no difference in the strength , a conjecture which was proved afterwards to be correct .
I have not given here an explanation of these or the other tables , as it is hoped they will be understood by inspection .
The forms of the fiactures of many of the pillars , particularly those in Tables I. and II .
, are given in conjunction with the drawings of the pillars in Plate XV .
The position of the neutral line was sought for with great care , and is inserted in the tables wherever it could be obtained with any degree of certainty .
Conclusions from the First and Second 7ables .
5 .
The experiments in the second table were undertaken after those in the first were nearly completed ; for I found , by a trial or two , that a pillar with flat ends was much stronger than one of the same diarmeter and length with its ends rounded .
To ascertain , therefore , the ratio of the strength in the two cases , and whether it was constant under all changes of dimension , it became necessary to make the second table to include most of the lengths and diameters of the former ; and , indeed , other considerations made me reduce the lengths in it , till the pillars broke by cruslhing without flexure .
Taking , then , the mean breaking weights of pillars from the same models in the first two tables of experiments , and neglecting the small differences in the measured diameters of the castings , we have the ratio of the strengths as in the following abstract .
Abstract from TABLES I. and II .
, &amp;c. Pillars with rounded ends .
Pillars with flat ends .
Ratio of breaking weights of pillars ... ... ... .-------with rounded and MeandiaNumber of Breaing Mean diaNumber of Brek Weight which flat ends , the diaMean diaof Breaking Length .
meter .
meter experiments we .
mei would crush meters being consideduced from .
deduced from .
g the cylinder .
dered equal .
inches .
inch .
lbs. inch .
lbs. lbs. 60'5 -50 2 143 '51 2 487 ' 22430 1 : 3-40560.5 '77 2 780 '77 2 2456 51130 1:3148 s 605 '99 2 1902 '997 3 6238 85721 1 3-279 c 60'5 1-29 2 5707 1*29 3 16064 143508 1:2-815 4 60-5 1 52 2 10861 1-56 3 28962 209869 1 : 2667 &gt; ? ?
30-25 '50 3 539 '50 1 1662 21559 1:3083 c , 30-25 '77 1 2726 '77 3 8811 51130 1 : 3232 30-25 -99 2 6105 1 01 4 20310 87971 1 : 3326 ? ?
15-125 -50 3 1904 -51 3 6764 22430 1 : 3552J _ 20-166 '767 2 6602 -777 3 15581 52064 1 : 23600 20-166 1-01 2 15737 1-022 2 31804 90074 1:2-0210 15-125 -76 3 9223 -775 4 21059 51797 1:2-2833 7-5625 '497 3 5262 -50 3 11255 21559 1 : 21389 15-125 -99 3 19752 1'000 2 40250 86238 1 : 2-0378 10-083 -76 3 17506 -768 4 25923 50865 1 : 1-4808 7-5625 '77 2 22948 '777 3 32007 52064 1 : 1-3947 3'7812 '50 2 15107 '52 3 24616 23319 1 : 16294 column , that the ratio of the strengths of pillars , with rounded and with flat ends , is nearly constant for all pillars , from those which are the longest compared with their diameters , to those whose length is about thirty times the diameter ; which is the case with the pillars 304 inches long , and 101 inch diameter ; and also those of 15'125 inches long and '51 inch diameter .
In all pillars , therefore , whose length is thirty times the diameter , or upwards , the strength of those with flat ends seems to be about three times as great as the strength of those of the same dimensions with rounded ends , the mean ratio being 1 : 3'167 .
In the pillars shorter than thirty times the diameter , the ratio is not constant : and it will be seen that it decreases when the length compared with the diameter decreases .
This uniformity in all pillars down to a certain length , compared with their diameter , and the departure from it in shorter pillars , were matters which for a considerable time arrested my attention .
Experinments were made upon wood and wrought iron , and the rlesults were in accordance with these ; but the reason of the deviation was first made evident by the experiments upon wrought iron , from which it appeared that short specimens of that metal becamne slightly crushed , and perceptibly changed in their form , by a weight which was nearly the same as that which produced a falling off in the strength of a long pillar of the same diameter .
Thus it was found that a weight of 22679 lbs. produced a very obvious change in the form of a shot cylnder , 015 in ch diameter as will be seen from the experiments further on ) : and a pillar of that diameter and 30 .
inches long , with the ends flat , exhibited a great falling off in strength .
It was evident , therefore , that incipient crushing or derangement of the parts was the cause of the change .
Cast iron does not often admit of such an alteration in its form as to be perceptible to the eye ; but its strength begins to fall off with weights which differ not widely from those which deranged the form , and reduced the strength of wrought iron .
I have , therefore , made the preceding abstract to include , in addition to the mean breaking weights from the first and second tables , a column containing the weights which would crush , without flexure , short cylinders of cast iron ( the Low Moor , No. 3 .
) of the same diameters as the pillars with flat ends .
These crushing weights were obtained from the experiments further on ( Art .
55 .
) , in which it was shown that prisns of this iron , perfectly bedded at the ends , and so short as not to be bent with the pressure , were crushed with a force of about forty-nine tons per square inch .
The mean of my experiments gave 86238 Ibs .
for the crushing force of a cylinder one inch diameter ; and the resistance to crushing , in the cylinders of other diameters , was calculated from this , on the supposition that the resistance is as the area of the section ; which supposition was proved to be true in my Report on the strength of hot and cold blast iron* .
7 .
It will be seen at a glance from the preceding abstract , what portion of its crushling strength each pillar required to break it , whether the ends were rounded or flat .
The pillars with flat ends , whose length was the least number of times the diameter ( the ratio of the strengths of the pillars with rounded and with flat ends being uniform ) , required 2430 and 87971 of the crushing weights to break them .
The lengths of these pillars were thirty times their diameters , which were *51 and 101 inch respectively ; but the pillars *777 inch diameter , and 20A166 inches long , ( the length being twenty-four times the diameter ) , were broken with 52064 of the crushing weight .
These last pillars , which were broken with a little less than onethird of the crushing weight , show ( l themselves injured by the compression , since they gave a ratio less than the mean .
On the other hand , the pillars just mentioned as '51 inch diameter were broken with from one-fourth to one-third of the crushing weight ; and those Ol01 inch diameter with somewhat less than one-fourth , neither of these being injured by the pressure .
We may , therefore , conclude that about onefourth of the crushing weight is the greatest load which a cast iron pillar , flat at the ends , will bear without producing a crushing or derangement of the materials , which would lessen its breaking weight ; and that the length of such a pillar should be thirty times the diameter , or upwards .
Pillars whose length is less than in this proportion , give the ratio of the strengths of those with rounded and with flat ends , from 1 : 3 , down to 1 : 12 , or less , according as we reduce the number of tinmes which the length exceeds the diamleter , as will be seen by the abstract .
A circumstance may be mentioned , which , perhaps , tended a little to equalise the strengths of the shortest pillars mentioned above .
It became necessary to render those which were rounded at the ends more flat there , than if the ends had been hemispheres ; whilst in the experiments upon pillars , whose length was greater with respect to the diameter than these , the ends were more prominent than in the hemispherical form .
This change became necessary on account of the splitting of the ends in the short pillars ; it having been found that the pillars whose diameter was one-thirteenth of the length or upwards , with rounded ends , failed in many instances by the ends becoming split .
In these cases a portion of the rounded end of the pillar formed the base of a cone , whose vertex was in or near the axis of the pillar .
This cone acting as a wedge whose sides were in the angle of least resistance , and having its vertex sharp , split and cut up the sides of the pillar , of which it folrmed a part* .
All the pillars in the preceding abstract became bent , and broke in the middle , as if the breaking weight had been transverse .
This was the case with all the pillars tried , till those of -52 inch diameter became only two inches long , the length being less than four times the diameter .
Sorne of the short pillars , it will be noticed , bore more than the average crushing weight , arising from these having been cast very small , and therefore being harder and resisting with more than the average strength of the metal .
i For more detailed information on the crushing of short masses of cast iron , and the interesting forms of its fractures , see the Report in the volume of the British Association before referred to .
389 We see , therefore , that when pillars of cast iron , with flat ends , are shorter than about thirty times the diameter , the ratio of the strength of those of the same dimensions , with flat and with rounded ends , is very variable .
But when the length is thirty times the diameter , or more , up to one hundred and twenty-one times , we have shown it to be nearly constant ; and we may infer that the ratio would be constant , however greatly the length might exceed the diameter ; since the uniformity seems to depend upon the circumstance that the weight necessary to break these long pillars by flexure , is not sufficiently great to produce much compression in them , and therefore does not alter sensibly the position of their neutral line .
In this respect , cast iron has the advantage of most other metals , on account of its great power of resisting crushing , which is , on the average , from six to seven times as much as that necessary to tear it asunder* .
Wrought iron , on the other band , does not require nearly so much force to crush it , so as in a great measure to destroy its utility t , as is required to tear it asunder ( Art .
60 .
) ; and therefore the ratio of the strength of equal pillars , with rounded and flat ends , becomes variable somewhat sooner in it than in cast iron , though the comparative results are the same in long pillars .
The experiments on timber , and on steel , so far as they have gone , give the same results with respect to long pillars .
9 .
Confining our researches , for the present , exclusively to what may be termed long cast-iron pillars , or those whose length , in pillars with flat ends , is thirty times the diameter , or upvards , since these are but little compressed by the breaking weight , we shall obtain some interesting properties .
These properties , doubtless , would be common to all rigid bodies , where the length , compared with the diameter , was so great that the breaking weight was not sufficient to diminish sensibly the power of the body to resist a transverse force .
Till recently , in all inquiries respecting the strength of imaterials , bodies have been assumed to be incompressible ; and if they really were so , the constant ratio above-mentioned , and some other properties , which I shall deduce from the results of the experiments upon cast-iron pillars , would be found in them , wxhatever might be their length , compared with their diameter .
Some descripti'ons of stone may perhaps approach much nearer than cast iron to this state of ultimate incompressibility , compared with their power of resisting tension .
Strength of Long Pillars of Cast Iron .
10 .
We have seen , that in all long pillars , where the dimensions are the same , the resistance to crushing by flexure is about three times as great when the ends are flat as when they are rounded .
This general agreement , and the near approach in point of strength , in pillars with flat ends , to those of the same diameter and of half the length with rounded ends ( as may be seen by comparing the preceding abstract ) , led me to conceive , that if the pillars with flat ends were more firmly fixed at the extre* Transactions of the British Association , vol. vi .
, before referred to .
t The valuable experiments of Mr. BARLOW refer to a further destruction of the material .
390 mities , their strength would be increased , so that they would bear the same as pillars of equal diameter and of half the length with the ends rounded .
To ascertain whether this would be the case , experiments were made upon pillars 60O inches long , from the same models as were made in Tables I. and II .
, but with this addition , the pillars were now cast with strong discs upon their ends , of about double the diameter of the pillar .
These discs were turned flat and perpendicular to the axis , and therefore rendered the ends of the pillar , when placed between the parallel crushing surfaces , perfectly imnmoveable .
The results of these experiments are given in Table IV .
11 .
Comparing the mean breaking weights in Table IV .
with those of the corresponding pillars in Table II .
, we shall find that pillars with discs on the ends are somewhat stronger than those without them .
I will select the mean results from the second and fourth Tables , from models of the same diameter and length ; and from the first Table , those from pillars of the same diameter and half the length ; affixing to each result a number in a parenthesis , indicating the number of experiments the mean has been derived from .
Abstract from TABLES I. , II .
, and IV .
Pillars with discs on the ends Pillars with their ends turned flat .
turned flat Pillars with the ends rounded .
Length 601 inches .
( Table II .
) Length 60inches .
( Table IV .
) Length 30T inches ( Table I. ) Diameter .
Breaking weight .
Diameter .
Breaking weight .
Diameter .
Breaking weight .
inch .
lbs. inch .
lbs. inch .
lbs. *51 487 ( 2 ) -50 539 ( 3 *77 2456 ( 2 ) -775 2719 ( 2 ) 77 2726 ( ) ?
997 6238 ( 3 1.00 6830 ( 4 ) 99 6105 2 ) 1'29 16064 ( 3 1.28 16369 ( 2 ) 1,29 17235 1.56 896 3 1-53 30789 ( 2 ) 1.52 32531 3 12 .
These results are , as we see , the means from a considerable number of experiments , agreeing moderately well among themselves : they show , as has been observed before , that the discs give the pillar a small increase of strength above that of a pillar with flat ends only ; and the approach to equality between the strengths of the pillars with discs , and of those of the same diameter , and half the length with rounded ends , is perhaps as near as can be expected in experiments of this nature .
Hence we may conclude , that a long uniform cast-iron pillar , with its ends firmly fixed ( whether by means of discs or otherwise ) , has the same power to resist breaking , as a pillar of the same diameter and half the length with the ends rounded , or turned so that the force would pass through the axis .
It may here be observed , that , in comparing the strengths , I have , as before , neglected the slight differences of diameter in pillars from the same model , and supposed them all of equal diameter .
13 .
Of the general conclusion arrived at above , I would offer the following as an explanation .
Suppose a long uniform bar , or bolt of cast iron , were bent by a pressure at its ends , so as to take the form Abed of B ; where all the curves AAb c , cd e , of B , separated by the straight line Ace B , would be equal , since the bar was supposed to be uniform .
The curve having taken this form , suppose the points b and f to be rendered immovable by some firm fixings at those points .
This done , it is evident we may remove the parts near to A and B , without at all altering the curve bc def of the part of the pillar c between b and f , and consider only that part .
The part b f , which alone we shall have to consider , will be equally bent at all the points b , d , f. The parts c and e , too , are points of contrary flexure ; consequently the pillar is not bent in them .
These points are unconstrained , except by the pressure which forces them together ; and the pillar might be reduced to any degree in them , provided they were not crushed or detruded by the compressing force .
These points may then be conceived as acting like the rounded ends in the pillars of Table I. ; and the part cde of the pillar , with its ends c and e supposed to be rounded , will be bearing the same weight as the whole pillar bc def , of double the length , with its ends bf firmly fixed .
14 .
The theory of the strength of pillars , as given by EULER and LAB GRANGE* , and aftelrwards pursued by POISSON and others , furnishes us with little information upon these subjects .
According to that theory , the strength is inversely as the square of the length ; or a pillar , half the length of another of the same diameter , would have four times the strength .
The results of my experiments give the strength nearer to three times , as a general value for the half length in cast iron .
It is , however , very variable , as will be seen by the first and second tables , unless we restrict ourselves to the longer columns .
The strength is much influenced , as has been previously observed , by the quantity of compression which the pillar sustains ; and , consequently , by the position of the neutral line when the pillar is bent .
The strengths , too , are different in their definition in the two cases .
In the theory of EULER , the strength is estimated by the greatest weight which a pillar would bear without flexure ; whilst in the present case , the estimate is formed upon the weight which would break the pillar by flexure .
I have sought , on many occasions , but without success , to determine experimentally some fixed point , according to the definition of the continental theory .
So far as I can see , flexure usually commences with very small weights , such as could be of little use to load pillars with in practice .
It seems to be produced by weights much smaller than are sufficient to render it capable of being measured .
I am , therefore , doubtful whether such a fixed point will ever be obtained , if indeed it exists .
With respect to the conclusions of some writers , that flexure does not take place with less than about half the breaking weight , this , as is evident from my experiments , taken in general , could only mean large and palpable flexure ; and it is not improbable that the writers were in some degree deceived from their having generally used specimens thicker , compared with their lengths , than have usually been employed in this paper .
Some results of the continental theorywe shall , however , find of great service further on .
Pillars with one End rounded and the other flat .
15 .
In the first and second Tables we have the strength of cast-iron pillars , with both their ends rounded , and both flat ; we will now refer to Table V. , which contains the results of experiments to ascertain the strength of pillars with one end rounded and the other flat , and in some cases with discs ; these results being placed between those from the pillars , with round and flat ends , as taken , with some additions , from the first and second Tables .
It has been shown , that the strength of pillars , with both their ends fixed , is to the strength , when the ends are rounded , in a constant ratio .
It seems probable , therefore , that the strength , when one end is fixed and the other rounded , must be related to the results of the other cases by some very simple law .
Taking the mean results from Table V. and other places , and neglecting the slight differences in diameter among castings from the same model , we have as below : Breaking weights in lbs. Both ends rounded ... ... ... ... ... ... . .
143 3017 7009 7009 16493 One end rounded and one flat , or with a disc .
256* 6278 '13499 13565.33557 Both ends flat , or with discs ... ... ... ... ... ..487 9007 0310 22475 16 .
The pillars in each vertical column in this abstract , are of the same length and diameter ; the strengths , therefore , in the three different cases , reading downwards , are as 1 , 2 , 3 , nearly , the middle being an arithmetical mean between the other two .
It is shown , moreover , by the experiments further on , upon timber , wrought iron , and steel , that the strength of a pillar with one end round and the other flat , is always an arithmetical mean between the strengths of pillars of the same dimensions with both ends rounded and both fat , however the ratio of the strength of these may varyt .
We see , too , by the first and second columns in the preceding abstract from Table IV .
, which contain the ratio of the strengths in pillars with discs , that their results are in accordance with the rest .
From these facts , as well as from other evidence not adduced , I conclude that the above mean would always take place , however the fixed ends were fastened .
17 .
The case of pillars fixed at one end and moveable at the other , is not without its interest in practice ; as piston rods of steam-engines and the rods of force-pumps are instances of the use of materials in that form , it being the same thing in effect whether the pillar be rounded at the end , or turn upon a joint there .
Other instances may often be seen in practice ; as where pillars are firmly based at the bottom , but through imperfect fixings , or some other cause , are capable of moving at the top .
It applies , too , if I am not mistaken , to the experiments upon timber by GIRARD * , whose pillars rested upon a flat surface at the bottom , but turned with a joint at the top , or were moveable there in the direction of the length of the lever .
But when the pillars were so placed as to have their least dimensions in that direction , they broke with the mean weight , as above defined , instead of the full weight , which they were expected to bear , as will be evident from a careful inspection of M. GIRARD'S plates attached to the work .
18 .
In all the previous experiments , where the pillars were uniform and the same at both ends , the fracture took place in the middle , or very near to it .
This , as might be expected , was not the case where one end was flat and the other rounded ; in that case the pillar broke at some point nearer to the rounded end , the piece broken off being always a little more than one-third of the whole length .
The excess , however , above one-third was so small , that I conceive this division would be absolutely that to which the material would tend , provided it were incompressible , and if the pillar were better fixed at the flat end , and more free to move at the rounded one .
With this view , two pillars , each thirty inches and a quarter long , and one inch in diameter , were formed with broad discs on one end , and rendered as nearly pointed at the other as was consistent with their bearing the pressure .
The result of the experiment upon these was this : the part broken off from one was 10-42 inches , and from the other 10-30 , whilst the one-third part of the whole length was 10'08 inches .
I was led to conceive the division above ( one-third ) not improbable , since it had been shown before , that a pillar with flat ends , or with discs upon them , bore the same as a pillar of half the length with rounded ends ; and as the pillars above had a disc on one end , and were rounded at the other , they might divide themselves so that the part with the flat end might be twice as long as that with the rounded one .
The results of my experiments on wrought iron , which is the only material besides cast iron in which I have examined the matter , give the division so as to make the piece , near to the rounded end , more in excess above one-third of the whole length than in cast iron ; the experiments , however , were made upon pillars too short to try the matter quite satisfactorily .
This deviation , if the above views are correct , may perhaps be attributed to the greater compressibility of wrought iron than of cast , as compared with their powers of resisting tension .
A better material to try it would be stone .
19 .
We have now seen , that when uniform pillars are so placed that one end is fixed and the other moveable , the point of greatest strain is at one-third , or a little more , of the whole length of the pillar from the moveable end .
We might , therefore , economise the metal by increasing the thickness in that point , and reducing the strength as we recede both ways from it .
Pillars enlarged in the Middle .
20 .
It has been stated above , that long pillars of uniform thickness , with both ends alike , always break first , in or near to the middle : this was the case even when they had discs upon the ends , to give the utmost firmness to their fixings .
It seemed evident , then , that the pillar was always too weak there .
I felt , therefore , desirous to ascertain the effect of strengthening the middle of pillars ; but as the ends could not be reduced at pleasure , since they would be crushed without bending , in the manner of our shortest pillars in Table II .
, I increased the diameter in the middle , leaving that of the ends the same ; and conceiving it best to make the experiments upon pillars whose form was as simple as possible , models were made of the form of two frustums of cones , the bases of which met in the middle of the pillar , the end diameter being 1 inch , and the middle 1 , 12 , 1- , 2 inches .
TIhe sides , therefore , were straight , and regularly tapering from the middle to the two ends .
Tables VI .
and VII .
contain the results of the experiments , the pillars being all ofn the first of these Tables , the extreme ends of the pillars were rounded , that the pressure might be through the axis ; and in the second , the ends had large discs upon them , turned flat .
Table VI .
shows , that in all the pillars with rounded ends , those with increased middles were stronger than uniform pillars of the same weight , thincrease being about one-seventh of the weight borne by the former .
In the pillars with discs , Table VII .
, those with the middle but little increased , had no advantage , with regard to strength , over the uniform ones .
But the pillars with the middle diameter half as great again as the end ones , bore from one-eighth to oneninth more than uniform pillars of the same weight with discs upon the ends .
&gt; Strength of long Pillars as dependent upon their Dimensions .
21 .
We shall now investigate the relative strengths of long pillars , as influenced by variations in their diameter and length .
From the theory of EULER , it would appear that the power of a pillar to resist incipient flexure is directly as the fourth power of the diameter , and inversely as the square of the length .
The inquiry in this paper is with respect to the resistance of pillars to fracture ; and , as has been mentioned before , I have not been able to find the point to which EULER'S computations refer .
His measures of the strength , however , seem not very widely different from those which apply to the breaking point .
Strength as dependent upon the Diameter .
22 .
I shall first ascertain , from the first and second Tables , the power n of the diameter to which the strength is proportional , in pillars of the same length .
Comparing the resistances of the pillars whose diameters were.50 and 1765 inches , and the length 601 inches , in Table I. , we find that the former was broken with 143 lbs. , and the latter with 15560 .
3E2 395 Since 1i765 3-530 '50 1 we have 1 " 3'53n : 143 : 15560 , and 3.53n 15560 ; 143 whence 15560 Log. 143 n ---3'718 .
Log. 3*53 I insert below values of n , thus obtained , from the first and second Tables ; and from the square pillars and those with discs , in Tables II1 .
and IV .
, beginning with long pillars with their ends rounded .
23 .
In Table I. , comparing the strength of a pillar of -50 in .
diameter , and 60 ' ins .
long , with that of '77 in .
diameter , gives n= 3'928 -50 in .
diameter , and 60 } ins .
long , with that of '99 in .
diameter gives n= 3'788 -50 in .
diameter , and 60ins .
long , with that of 1'29 in .
diameter , gives n= 3'889 -50 in .
diameter , and 60 } ins .
long , with that of 1-52 in .
diameter , gives n= 3'894 -50 in .
diameter , and 604 ins .
long , with that of 1-765 in .
diameter , gives n= 3'718 -50 in .
diameter , and 604 ins .
long , with that of 1'94 in .
diameter , gives n= 3'741 '77 in .
diameter , and 60ins .
long , with that of 1'9765 in .
diameter , gives n= 3'609 *77 in .
diameter , and 604 ins .
long , with that of 1194 in .
diameter , gives n= 3'654 750 in .
diameter , and 304 ins .
long , with that of 1'29 in .
diameter , gives n= 3'656 '50 in .
diameter , and 304 ins .
long , witti that of 1-52 in .
diameter , gives n= 3'687 '77 in .
diameter , and 301 ins .
long , with that of 1'52 in .
diameter , gives n= 3-637 -99 in .
diameter , and 304 ins .
long , with that of 1'52 in .
diameter , gives n= 3'902 950 in .
diameter , and 153 ins .
long , with that of -76 in .
diameter , gives n= 3'768 -50 in .
diameter , and 154 ins .
long , with that of '99 in .
diameter , gives n= 3-425 Mean value of n= 3'736 In this comparison I have excluded all pillars , the length of which was not greater than about fifteen times the diameter .
In pillars shorter than in about this proportion , a great change doubtless takes place by degrees in their elasticity , and their strength is much reduced .
This remark is intended to be understood only of pillars with rounded ends : in those with flat ends , the samrne change , we have seen , begins to take place when the length of the pillar is less than thirty times the diameter .
As this change in the elasticity of pillars must be produced by the same weight ( the diameter being supposed constant ) , the relative strength of pillars with rounded and with flat ends , to resist the change , must depend upon the laws that regulate their ultimate strength ; and we have seen that a pillar with rounded ends will bear the same as one of double the length with flat ones .
In the experiments , Table III .
, on square bars 396 inches long , rounded at the ends , comparing togethler the strengthls of the pillars whose sides are '7716 and 1'0275 , gives n= 3'9830 ; '7716 and 1'540 , gives n= 3-6789 ; 1'0275 and 1'540 , gives n= 3'4698 ; Mean n= 3'7105 .
24 .
From the two preceding inquiries , as to the value of n , we see , that whether the pillar be circular or square , the power of the thickness to which the strength is proportional is nearly the same .
We may therefore conclude , that the same value of n would be obtained from any other form of section .
In the values obtained from the circular pillars , which include specimens varying in length from twenty times the diameter to one hundred and twenty times , the values of n do not differ so widely , but that they may be considered as constant .
They in no case rise so high as four , the power according to EULER ; but in pillars , so long with respect to their diameter that the breaking weight is very small , they approach nearly to it .
We may , therefore , say that four is the power to which they would arrive if the breaking weight was so small that it produced no compression in the pillar .
Long Pillars with the Ends flat .
25 .
We will determine the values of n , the power of the diameter to which the strength is proportional , in these pillars , in the same manner as was done in the former .
In the cylinders 607 inches long ( Table II .
) , comparing th of that of *77 inch diameter with that of -997 , give : n= 3,606 ; .77 inch diameter with that of 1P29 , gives n=3 641 ; '77 inch diameter with that of 1P56 , gives n= 3'495 ; '997 inch diameter with that of 1-29 , gives n= 3-670 ; -997 inch diameter with that of 1P56 , gives n= 3'428 ; Mean n= 3.568* .
The pillars of the same length , with discs upon the ends , in Table IV .
, if we make the same comparison of strength , give as below : Comparing that of *51 inch diameter with that of 1-00 diameter , we have n= 3'922 ; *51 inch diameter with that of 1-28 diameter , we have n= 3'820 ; *51 inch diameter with that of 1-53 diameter , we have n3-775 ; '775 inch diameter with that of 1l53 diameter , we have n= 3'568 ; *775 inch diameter with that of 1-28 dianeter , we have n= 3'578 ; 1-00 inch diameter with that of 1-53 diameter , we have n= 3'412 ; Mean = 3'679 .
The experiments in Table II .
, upon pillars shorter than these , are not in sufficient variety to enable us satisfactorily to pursue the comparison ; and if we attempted one between the strengths of the short pillars in that Table , as , for instance , those of oneor one-fourthird , thor one-fourth of the sabove , we should find the value of n very low , on account of the crippling or other great change which we have seen ( Art .
6 ) that pillars sustain by the weight , when they are made shorter than abouhirty times ththirty times t diameter .
Strength as dependent on Length .
26 .
In order to obtain the power of the length , to which the strength of pillars of the same diameter is proportional , considering this s as some inverse power x , we will determine its value from the first and second Tables , beginning with Table I. , in which the lengths vary as widely as eight and sixteen to one ; and we shall take the values from the whole range in that Table .
27 .
Here we find that a pillar of half an inch diameter and 60'5 inches long , was broken with 143 lbs. , and one of the same diameter and 3'78125 inches , the one-sixteenth of the former length , required 15107 lbs. to break it .
Proceeding in the manner adopted before , we have , Lx : 16x : 143 : 15107 .
Whence we obtain x= 1'6807 .
In this comparison I shall assume all the pillars which were cast from the same models , to be equally thick , neglecting occasional slight differences in their diameters .
28 .
Pillars with their Ends rounded .
Diameters of cylinders Lengths of which the Yalues of .
Mean values of .
considered equal .
strengths are compared .
inch .
inch .
inches .
inches .
*50 -50 60-5 and 30-25 1-9141 -50 -50 60-5 15 151867 1 I. 6 '50 -497 60-5 7-5625 1-734 1 f799 '50 '50 605 3-78125 1.681 J *77 -77 60-5 30'25 1-805 ) 1 *77 '76 60-5 15-125 1-782 | '77 -77 60-5 7-5625 1-626 &gt; 17032 -77 '76 30-25 15-125 1-758 '77 -77 30-25 7-5625 1-537J *99 -99 60'5 30-25 16821 ' 99 -99 60S-5 15-12S 1-688 * 85 1-29 1'29 60-5 30-25 1-595 15595 '1'52 1'52 60-5 30'25 1-583 1-583 r_ * i.-miln n -------- , - , . ? ? ? ?
Pillars with their Ends fat .
Diameters considered Lengths of which the Values of x. Mean value of X equal .
strengths are compared .
inch .
inch .
inches .
inches .
*77 -77 60-5 and 30*25 1-843 *77 -777 60-5 20-1666 1-682 1625 *77 '775 60.5 15-125 1-550 ( *77 -785 60.5 12 100 1-424J.997 1-01 60.5 30-25 1-691 1.587 ?
997 1-022 60*5 20-1666 1-483 30 .
The value of x , as exhibited in the preceding abstracts from Tables I. and II .
, is not constant .
Taking pillars of any particular diameter , we see that x decreases according as we decrease the length of the pillar ; and consequently increase the pressure necessary to break it .
Thus , in pillars with rounded ends , taking those of '50 inch diameter , and comparing the strengths of those of 60inches long with those of 30-25 , 15-125 , 7-5625 , 3-78125 inches , we have x= 1P914 , 1-867 , 1-734 , 1-681 .
In pillars with flat ends , taking those of which the diameter is -77 inch nearly , and comparing those of 601 inches long with those of 30-25 , 20-1666 , 15*125 , 12-1 inches , we obtain successively , x= 1843 , 1'682 , 1'550 , 1*424 .
This decrease in the value of x also takes place in the comparison between the different lengths of the pillars -99 inch diameter ; and it is evident that it would do so in all cases where the diameter was the same , whatever that diameter might be .
The falling off in the strength of pillars , whose length is less than twenty-five to thirty times the diameter , in those with flat ends ; and about one-half that length , in those with rounded ends , as before observed , is not the sole cause of this diminution in the value of x , though it is a part of it .
Indeed , we may perhaps say that both depend in some degree upon the same cause , the compressibility of the material .
The increasing weight alters the position of the neutral line in the pillar , and consequently lessens the part subjected to tension .
By this means the flexibility of the pillar is increased , and its resistance decreased ; because this resistance depends upon the transverse strength and the power to resist flexure .
The position of the neutral line , on which this variation of the strength depends , has been obtained in several of the cases of Tables I. and II .
; and indeed many others .
31 .
In the abstract from Table I. , just given , and in which the lengths vary as widely as sixteen to one , we find the highest value of x= 1*914 , and the lowest = 1537 ; and the mean from all the values = 1'7117 .
The abstract from the pillars with flat ends , in Table II .
, is not pursued to the same extent as the former .
In it , the highest and lowest values of x are 1-843 and 1P424 ; the mean from the whole of these being 1-612 .
32 .
The pillars compared above comprise a much greater range of lengths than those in general use ; and as the shortest pillars do not come withiin the simlple rules applicable to the longer ones , I shall adopt 1-7 in the future calculations as an approximate value of x , applicable to most of the pillars used in practice .
399 We see from the preceding abstracts , that the values of x never rise so high as two , but approach toward it accordingly as the breaking weight becomes smaller .
In the pillars *77 inch diameter and 60 } inches long , where the length is seventyeight times the diameter , the value of x is 1'805 in one case , and 1[843 in another ; and in the pillars '50 inch diameter , and 601 inches long , the length being one hundred and twenty-one times the diameter , the value of x is 1'914 .
These facts , and the regular increase in the value of x according as the breaking weight is diminished , show that 2 is the value to which x would approximate if the breaking weight were infinitely small , or the body perfectly incompressible .
Computations of the Strength of Long Pillars by means of the preceding Numerical Results .
34 .
We have found that the strength of cast-iron pillars , whose diameter is the same , is inversely as the 1*7th power of the length nearly ; and where the diameter differs , the strength varies according to a power of the diameter which is nearly constant .
This , in cylindrical pillars , whose ends are rounded , we have found to be 3*736 , and in those with flat ends 3'568 , the pillars with discs upon the ends giving 3'679 .
The earlier experiments gave 3*76 for the pillars rounded at the ends , and 3*55 for the flat ones ; and these numbers will be retained in the following computations , whatever the form of the section may be ; it having been shown that the power was nearly the same , whether the pillars were round or square .
35 .
As a convenient unit of comparison , we may determine the strength of a solid pillar whose length is one foot and diameter one inch , as obtained by calculation from most of the experiments in the first and second Tables .
Putting then d for the diameter of a given pillar in inches , I for its length in feet , w for its breaking weight , and x for that of a pillar one inch diameter and one foot long , we have - , as a comparative measure of the strengthl in pillars with rounded ends .
Whence d376 1376.* .
X =376 X for pillars with their ends rounded .
In the same manner , for pillars with their ends flat , x= d3.55 .
400 The results from the pillars with rounded ends in Table I. are as follow : Diameter Breaking Mean Error from using of pillar .
Length .
weight .
value of x. the mean , in parts of the value of x. .
- ... I,.- , inches .
feet .
60'5 = 5-04166 30-25 = 2-5208 15-125 = 1-2604 7-5625 = -6302 60-5 = 5-0416 30-25 15-125 10-083 60'5 30-25 = 2-5208 = 1-2604 = -84028 = 5-04166 = 2-5208 20-1666 = 1*6805 15-125 = 1-2604 60o5 = 5-04166 = 2-5208 = 5-0416 = 2'5208 = 5-0416 = 5-0416 lbs. 143 lbs. 30309 35162 inch .
'50 *50 *50 *497 *77 *77 *76.76 *99 '99 1'00 '99 1.29 1.29 1.5275 1'52 1*772 1.95 30-25 60*5 30-25 60'5 60-5 539 1904 5262 780 2726 9223 17506 1902 6105 15737 19752 5707 17235 10750 32531 16562 23551 1 9.8 1I 19'7 1_ 42 + -L 7'8 .
1 42 1 207 1 20'7 11 12 , + 10-7 1+ 1-~ __1 14'2 1+1 38'2 209 1 40'2 +1 3'8 38230 33263 32603 35070 38360 36547 30902 30529 36639 30401 34273 31857 34229 32447 30092 29901 0 4q .
11 x2 .
CO CO The enumeration above includes pillars varying in length from 121 times the diameter down to 15 times , or as 8 to 1 .
Through this range the greatest error arising from using the mean value of x here obtained is 7,6 or somewhat more than one-eighth .
We shall , therefore , take 33379 lbs. , or 14'9 tons , as a unit of measure for the strength of a solid uniform cast-iron pillar , of Low Moor iron , No. 3 , one foot long and one inch diameter , the ends of which are rounded .
36 .
For solid pillars of other dimensions with rounded ends , in the same material , taking the length in feet and the diameter in inches , and using the same notation as before , we have w the strength as below : d3'76 w= 33379 -syfor the strength in lbs. d3'76 w= 14'9 -i .
for the strength in tons .
This rule is only intended to apply to pillars whose length is fifteen times the diameter or upwards , and in the largest pillars , perhaps , not so low .
It appears probable that we should find a falling off in the strength of these when they are considerably more than fifteen times the diameter in length .
This reduction in the resistance of the thicker pillars , compared with what we are led to expect from the results of the smaller , arises , as appears to me , from the greater softness of the metal when cast in large masses than in small ; and this cause , I conceive , has some influence upon the strength of all cast-iron pillars .
37 .
To find x , the strength of a pillar one inch diameter and one foot long , in solid cylinders flat at the ends , from the mean results in Table II .
Using here the formula X -e d3'e 5 and restricting ouw selves to cylinders , hose length is upwards of twenty-five times the diameter , we have , The abstract above contains pillars varying in length from 25 to 121 times the diameter , and omitting the first , for the reason given in the note , the greatest error which has occurred by using the general mean 98922 lbs. , is.4th .
Hence we may adopt 98922 lbs. , or 44 16 tons , as the strength of a cast-iron pillar , one inch diameter and one foot long , flat at the ends ; this number , however , only to be applied to pillars whose length is 30 times the diameter , or upwards .
38 .
Retaining the same notation as before , we have , for solid cylindrical pillars of any given size , the lengths being as above : w= 98922 X d--5.7 , for the strength in lbs. d3.55 w= 44'16 X d-7- , for the strength in tons .
*I conceive this result to be too low , on account of the difficulty of having slender pillars quite straight .
It is , therefore , not included in the general mean value of x. Diameter of Length of pillar .
Breaking Valueof x. Mean value Error from using the mean , pillar .
weight .
of x. in parts of the value of x. inch .
inches .
feet .
lbs. lbs. *51 60-5 = 5*04166 487 83182* m.77 60'5 = 5-04166 2456 97175 *50 30-25 2-5208 1662 93735 *77 30-25 2-5208 8811 107300 *777 20-166 16805 15581 92229 + 1.997 60-5 = 5-0416 6238 98638 1i 1.01 30-25 = 2-5208 20310 94403 1-29 60'5 = 5-0416 16064 101770 156 60-5 = 5-0416 28962 93455 m.51 20-166 = 1-68055 3830 101063 00 1 *51 15-125 = 1-2604 6764 109445 9.4 402 Short Pillars .
39 .
The mean value 98922 Ibs.would be too great , if used for calculating , by the preceding formula , the strengths of solid round pillars , which are shorter than about 30 times the diameter .
I shall , therefore , insert below the values of x , from the shorter pillars in Table II .
; together with the particulars of the experiments from which they were derived .
Putting these in a line with the real breaking weights , and the number of tons per square inch , upon the section of the pillars , I will endeavour to show the cause of the falling off in strength .
No. of , , , .
Tons per square Valueof from experiDiameter .
Length of pillar .
117 *eri Dlameter T All those values of x are in excess of the part of the general mean they are compared with .
We might , therefore , obtain a rough approximation to the strength of short pillars by calculating , by the formula before used , the strength of the pillar ; and then taking such a portion of this calculated strength as the length of the pillar is of what it would have been if it were 30 times the diameter .
This easy rule would nearly in all cases give the strength somewhat under the real strength , as appears from the instances we have cited , and others we might take from the abstract .
More correct mode of estimating the strength .
41 .
We have seen , that when pillars are reduced in length beyond a certain extent , there is a reduction in their strength from that which is given by the rules for longer pillars ; and this falling off is nearly in proportion to the reduction in the length of the pillar .
Another mode of considering the matter will , it is conceived , throw some light upon the reason of the conclusion in the last article , and furnish better means of estimating the strengths .
The two last pillars , in the abstract last given ( Art .
39 .
) , broke by being crushed , all the others having been broken by flexure .
One broke with 48'1 tons per square inch ; and the other with 51-7 .
The mean result of the experiments , upon the resistance to crushing in this metal , gave 49 tons per square inch nearly ( Art .
55 .
) .
'rakinrg a mean between the results of such pillars in the abstract as have their length thie same number of times the diameter nearly , we have as below , in experiments 3 , 4 and 5 , a pressure of 19 16 tons answering to a value of x= 75370lbs .
; a mean from the experiments 6 , 7 , 8 and 9 , gives a pressure of 23*97 tons per inch , answering to a value of x 53770 lbs. ; and a mean between the results in the 10th and 11lth expe.riments , gives 34'9 for the pressure per inch , and the value of x= 32241 .
We therefore see that -9 of the crushing weight reduced the value of x from 23'97 34.9 98922 to 75370 ; 4397 reduced it to 53770 ; and 4reduced it to 32241 .
In other words , when the pressure is a little more than two-thirds of the crushing weight , the strength to bear flexure is reduced to less than one-third ; when the pressure is a little under half of the crushing weight , the value of x is somewhat more than half its first value ; and when the pressure is less , the reduction is diminished .
Comparing the weights , per square inch , with all the values of x , in the two columins , we find , amidst many anomalies , that the weights increase in some such proportion as the values of x decrease ; and the latter become very small where the pressure approaches to the crushing force .
Considering then the pillar as having two functions , one to support the weight and the other to resist flexure , it follows , that when the material is incormpressible ( supposing such to exist ) , or when the pressure necessary to break the pillar is very small , on account of the greatness of its length compared with its lateral dimensions , then the strength of the whole transverse section of the pillar will be employed in resisting flexure ; when the breaking pressure is one-half of what would 404 be required to crush the material , one-half only of the strength may be considered as available for resistance to flexure , whilst the other half is employed to resist crushing ; and when , through the shortness of the pillar , the breaking pressure is so great as to be nearly equal to the crushing force , we may consider that no part of the strength of the pillar is applied to resist flexure .
We may , therefore , separate these effects by taking , in imnagination , from the pillar ( by reducing its breadth , since the strength is as the breadth ) as much as would support the pressure , and consider the remainder as resisting flexure to the degrees indicated by the previous rules ( Arts .
36 and 38 .
) .
42 .
Suppose then c to be the force which would crush the pillar without flexure ; d the utmost pressure the pillar , as flexible , would bear to break it without being weakened by crushing ( Art .
6 .
) ; b the breaking weight , as calculated by Articles 36 and 38 .
; y the real breaking weight .
Then , supposing a portion of the pillar , equal to what would just be crushed by the pressure d , to be taken away , we have cd= crushing strength of the remaining part , and yd the weight actually laid upon it .
Whence yd the part of this y-d -e-y remaining portion of the pillar which has to resist crushing , .
' .
1d --y _ the part to sustain flexure .
But the strength of the pillar , if rectangular , may be supposed to be reduced by reducing either its breadth , or the calculated strength of the whole , to the degree indicated by the fraction last obtained .
In the circle this mode is not strictly applicable ; but we obtain a near approximation to the breaking weight , y , by reducing the calculated value of b in that proportion .
We have , therefore , c -d : c-y : : b:y , : .
cy -dy beby , bc whence Yb+ -d 43 .
It has been shown ( Art .
7 .
) that about one-fourth of the crushing weight was the utmost that a flexible pillar could be broken with , without the material being crushed , and the breaking weight reduced in consequence .
In the Table following , I will give the values of y , as deduced from all the pillars in c b_ c-ore concluding this subb c Article 39 , taking d= 4 ; and , therefore , y= b-+~ c Before concluding this subject , I may , however , observe here , that there is a falling off in the strength of pillars , in consequence of the pressure , through all stages , from the smallest force necessary to break them to the largest , though this diminution does not become so obvious as to need correcting for , till the breaking weight is about one-fourthi of the crushing weight .
This reduction in strength will be shown from a comparison of the strengths of the thicker with the more slender pillars , the length being the same , the falling off always increasing with the diameter .
This will be rendered evident from the decrease in the 405 values of n ( Art .
23 to 25 .
) .
We may also see its effects in the comparison between the strengths of pillars 1P29 diameter , and between those of 1*56 , in the abstract ( Art .
5 .
) .
In each of these cases the pillars were of the same diameter nearly ; but as the breaking weights were , from another cause , very different , the comparative strength was considerably lower than in the pillars of smaller diameter .
This falling off in the strength of large pillars of equal lengths arises principally from this circulmstance , that the strength to resist flexure increases as a power of the diameter somewhat under the fourth ; and the resistance to crushing only as the second .
Thus the pillars of greater diameter require a much greater load upon the same area of section to break them than those with smaller diameters .
The position of the neutral line is therefore different , and the area of the part submitted to tension reduced ; the pillar , therefore , becomes more flexible , and consequently weaker .
Some small part of this falling off in the strength of larger pillars might be accounted for by the softer quality of cast iron when cast in large masses ; but other materials exhibit the same diminution of strength , and the cause above is sufficient to account for it .
bc Tabulated results of the formula y=c Diameter Length of Breaking ' Calculated breaking of pillar .
pillar .
value of b. value of c. weight .
Clat from formula , .
- , b. Mr__ __fory .
inches .
inches .
lbs. lbs. lbs. lbs. -50 121 8327 21559 7195 7328 -50 10-083 11353 21559 8931 8872 -50 7'5625 18515 21559 11255 11508 '50 3-7812 60155 21559 17468 16992 -777 20-166 16713 52064 15581 15604 '775 15-125 27005 51797 21059 !
21241 -785 .
12'1 41300 53142 24287 27043 *768 10-083 52096 50865 25923 29363 -777 7-5625 88547 52064 32007 36130 1'022 20-1666 44218 90074 31804 35631 1'000 15-125 66746 86238 40250 43797 The formula for y gives results agreeing moderately well with the strengths of pillars decreasing in length , both here and in the experiments on hollow pillars , as will be seen further on ( Art .
52 .
) .
Similar Pillars .
dS'76 44 .
We have found the strengths of solid pillars of cast iron to vary as l-7 and 7 , i5 nearly ; when they are not shorter than from fifteen to thirty times the diameter , according as the ends are rounded or flat ( Art .
36 to 38 .
) .
In the research for these numbers I have been led to conclude , that , if the material had been incompressible , the 3-76 and 3*55 would each have become 4 , and the 1-7 d4 have been 2 ( Art .
24.33 .
) .
In that case the strength would have varied as - , which is 406 the ratio according to the theory of EULER .
In similar pillars the length is to the diameter in a constant ratio .
Calling the length n d , where n is a constant quantity , we have , in these different cases , the strength as d3'76 d3'55 d4 W xdl ndx d ' dd or , making the divisions , these become d2'06 d"l5 d2 In the first of these cases , the strength varies as a power of the diameter somewhat higher than the square ; in the second , somewhat lower ; and in the third , as the square .
We may therefore conclude , that in similar pillars the strength is nearly as the square of the diameter , or of any other linear dimension * : and as the area of the section is as the square of the diameter , the strength is nearly as the area of the transverse section .
The experiments in Tables I. and II .
give for the strength of similar pillars , powers of their linear dimensions as below .
Diameters of pillars Length of pillars Breaking weight Powers of the dimencompared .
compared .
of pillars .
sions .
inch .
inches .
lbs. er *'497 ; 7'5625 5262 1 .
08 XI *99 K 15*125 i 19752 9I S..76 15-125 9223 !
1-52 30-25 32531 1.819 4I 99 30-25 6105 |L 1.97 i 60.5 25403 57 'f 51 20-166 i 3830 S : 1-56 .
60'5 28962 '41 |2 -50 30-25 1662 , 11 i -'997 60-5 : 6238 19081 I ' -51 15-125 6764 1| 1'02 30'25 21844 .
'-50 10-083 8931 L 1-022 20-166 31804 1J In the Table above , the pillars compared were from models which were similar .
I have , therefore , neglected slight deviations from similarity in the castings from these models .
It appears then that the power of the lineal dimensions , according to which their strengths vary , is somewhat lower than the second .
45 .
If pillars be so formed as equally to resist being crushed ( as shown in Art .
6 .
) by the breaking weight , they will be similar .
We have seen , that when pillars require a force to break them by flexure , which exceeds a certain portion of the force which would crush them without flexure , the pillar sustains a considerable diminution in its power of resistance to fracture by flexure , in consequence of a partial crushing , or crippling of the material .
Suppose c d2 = the crushing force of the pillar ( d being its diameter ) , or that pressure which would cause rupture in it , if it were too short to break by flexure ; and nc d2 that part of this pressure , which is the utmost it would , as flexible , sustain without apparent crippling or crushing .
Then , since the strengths to resist fracture by flexure in pillars , with both ends rounded d3'76 d3'55 and both flat , were 33379 1-yj- , and 98922 11--7 respectively ( Art .
36 .
38 .
) , we have these quantities each equal to nc d2 , in the cases where short pillars are bearing the greatest weights they can sustain without any apparent crushing .
Whence , In pillars with rounded ends , 3*76 / S^S 7OX^ 176 / 33379\ '76 33379l=ncd2 ; .
1=(3--79 ) X d'7 ; In pillars with flat ends , d3'*55 9-r1.39 ?
r 1-7\ 1'755 98922 =ned2 ; '. .
1= 9c )X dn- ?
In the former of these cases , I varies somewhat faster than as the first power of the diameter , and in the second , somewhat slower ; the two showing that , in the case of pillars equally loaded to resist crushing by the weight , the length to the diameter will be nearly in a constant ratio , or the pillars must be similar .
46 .
Strength of long uniform Pillars , as dependent on their weight , the length being constant .
dm in general , the strength varies as -H ; where d and I are the diameter and length ; and m , n , the constants on which the strength depends .
If the length be the same , the strength varies as dm ; and since the weight q of the pillar is as the square 1m of the diameter or directly as the area , d oc q2 ; whence strength oc q 2 .
376 In pillars with rounded ends , m= 3*76. .
strength oc q2 oc ql88 .
3*55 In pillars with flat ends , m= 3'55.- .
strength oc q-T oc q'1775 .
These values were used in comparing the strengths of pillars of equal weight in Tables VI .
and VII .
, and in other places .
It will be borne in mind that this proposition applies only to pillars above a certain length , as in Articles 36 and 38 .
Hollow Pillars of Cast Iron , the Low Moor , No. 3 .
47 .
The preceding observations have been made upon solid pillars only , as these , on account of their greater silplicity , are not only more easy to be cast than hollow ones , but present fewer difficulties in experimenting upon , and in deducing conclusions .
Solid pillars may be made as small as may be desired , but this is not the case with hollow ones .
These require great care on the part of the moulder , and are consequently expensive .
The first hollow pillars used in these experiments were cast in a horizontal position , but it was found impracticable in this way to keep the core in the centre of the casting , and therefore several of the first pillars were defective , in having one side much thicker than the other , as well as in having many air bubbles in them .
These causes tended to diminish the strength of the pillars ; and will account for some results in Table VIII .
being lower than they otherwise would have been .
The circumstance of some of these pillars being much thicket on one side than the other , gave me an opportunity of observing what falling off in strength arose from that cause , which is one to which cast-iron pillars , as used in practice , mIust be very liable , though not so much so as in these small experimental castings .
I have therefore recorded these cases with the others , conceiving that they would be interesting to practical mnen ; and it is gratifying to find that a matter , which would seem to destroy all confidence in a pillar , does not produce a great reduction in the strength .
The cause seemed to be this : in almost every case , where pillars were much thicker , in the middle , on one side than the other , the thin side was that which was compressed : and as cast iron resists fracture by compression , with ( on the average ) about six times the force that it resists tension , the pillar seldonm gave way by compression , and therefore bore nearly as much as it would have done if of equal thickness on both sides .
If the thick side of the pillar had become that which was compressed , which , for what I can see , was as likely as otherwise , the strength would probably have been much decreased .
But the circumstance of the thin side having been so often the compressed one , must apparently have arisen from some general cause .
The difficulty of obtaining good uniform castings caused me to mention the circumstance to Mr. FAIRBAIRN , who gave orders that all the fuiture pillars should be cast vertically , and in dry sand ; and this , together with great care , produced good castings , though the core could not always be kept in the centre towards the middle of the pillar .
The greatest length of pillar to which the apparatus was adapted , was about seven feet six inches ; being a length half as much again as the greatest which was used in the experiments upon solid pillars .
As , in the former experiments , the five-feet pillars had , by a mistake in the coimmencement , been made half an inch too long , my anxiety to avoid error , however slight , caused me to have the lengthls of all the shiorter pillars exact subdivisions of the longer ; this entailed upon all the future experiments fractional lengths , which have been introduced into the hollow cylinders , in order to make them exact multiples of the former .
The length of the hollow pillarts was therefore made 7 feet 63 inches ; and it was iny wish ( which was always seconded by my liberal friend Mr. FAIRBAIRN ) that their diamtneters should be as varied as possible , both for theoretical purposes and the application of their results to practice .
There were difficulties , however , which militated against the execution of this desire .
The external diameter of the pillars could not be reduced at pleasure ; the internal must be large enough to admit a small hollow metallic tube , covered over with a thin stratum of loam , tied together with vegetable matter , the tube being perforated in every part with holes , and having its internal diameter large enough to allow the gas and steam to escape .
Through these causes , the smallest pillars which we could cast of the above length were 1 inch external diameter , and 1 internal .
On the other hand , the experiments were bounded , in the largeness of the diameters , only by the power necessary to break the pillars .
With Mr. FAIRBAIRN'S apparatus ( a single lever of which the greatest multiplier was 7 or 74 ) I had never conceived it safe to apply more pressure than about 18 tons ; and when an additional lever was added to this , the greatest weight used was 22tons .
This was in one of the experiments on hollow pillars .
Through these causes , the variation in the diameters of the longest hollow pillars was necessarily small , particularly in those with flat ends , of which the largest external diameter was about 24inches , this requiring upwards of 18 tons to break it .
In this , the length of the pillar was 40 times the external diameter , and therefore might be considered as representing only the longest and slenderest pillars used in practice .
In this matter , however , I was not without a resource , it having been discovered , in the early part of these inquiries , that the ratio of the strengths of solid pillars with rounded ends , to those with flat ones , was nearly constant , and about as one to three ; and this was the case with all pillars , from those of the greatest length , as compared with the diameter , down to a certain ratio of length , considerably shorter than we could command power to break , provided the ends of the pillar were flat .
It was therefore determined to pursue the experiments upon these hollow pillars ( putting solid hemispherical caps upon the ends ) , increasing the diameters by small degrees ( the length being the same ) , till I could break them no further .
In this way I was enabled to vary the external diameters from 1*74 to 3*36 inches the former having sunk with 5711 lbs. , and the latter with 50477 lbs. , or 22tons .
In the largest pillars of this series , experiments 15 and 16 .
, Table VIII .
, the length was reduced to twenty-seven times the diameter .
To pursue the matter further , it now became necessary to reduce the length of the pillars , because when that was done the internal diameter could be reduced , and thus the external might be cast so small that the strength of the metal in the pillar would be overcome by the power which I had at command .
Several of these pillars were , at Mr. FAIRBAIRN'S suggestion , turned outside and bored within , to make their dimensions as uniform as possible .
By adopting pillars of smaller size than before , I was now enabled to render them as short , compared with the diameters , as I wished , and thus obtain specimens of most of the varieties used in practice .
( See Table X. ) Tables VIII .
and IX .
contain the results of the experiments upon the longest hollow uniform cast-iron cylinders ; with an experiment or two in each , upon solid cylinders of the same length to comipare with them .
The cylinders were always cast 410 with great care , and were either perfectly straight , or they were rendered so in the lathe , where their ends were turned flat and parallel .
The pillars intended to be used with rounded ends , were made about two inches shorter than the rest , or 7 feet 4 inches long , in order to allow roomn for the rounded caps upon the ends , as before mentioned , the whole length of the pillar with caps being 7 feet 63 inches .
Description of the Tables .
48 .
The first column in the Trables on hollow cylinders contains thie number of the experiments only , and needs no explanation .
The second column , in giving the internal diameter of the pillar , gives it as calculated as well as measured .
This was done because there was great difficulty in measuring correctly the internal diameters , which were in some degree irregular .
The weight of a solid pillar of a certain diameter , and the given length , being known , from a mean of three experiments , and the weights of the hollow cylinders , together with their external diameters , being also known , from these data , the internal diameters of all the cylinders were computed ; and a mean between the calculated and the observed results was taken for the internal diameter .
This extra trouble was taken , as I was desirous of avoiding every possible inaccuracy in experiments which , from the expense and labour attached to them , were not likely to be soon repeated .
The third and fourth columns of Tables VIII .
and IX .
contain the deflections caused by different weights laid upon the pillar , and the weights themselves .
The fifth column contains what is denominated the breaking weight , but which means only that weight which was sufficient to overcome the greatest resistance of the pillar .
This was frequently attained when a long pillar had been bent through a space perhaps not exceeding its radius .
From the first experiment on long hollow pillars with rounded ends , it was evident that so little flexure of the pillar was necessary to overcome its greatest resistance ( and beyond this a smaller weight would have broken it ) .
that the elasticity of the pillars was very little injured by the pressure , if the weight was prevented from acting upon the pillar after it began to sink rapidly , through its greatest resistance being overcome .
By attending carefully to this point , the pillar was frequently saved uninjured for another experiment .
The small flexure that it had got was straightened , and the rounded ends being removed , the pillar , with its flat ends well bedded against the crushing surfaces , was broken , and the results in Table IX .
were recorded .
The sixth columns in Tables VIII .
and IX .
contain the strength of a solid pillar , 7 feet 6 ' inches long , and one inch diameter , as calculated on a theoretical supposition* mentioned further on .
This value is always calculated for the full length , 7 feet 6inches , though some of the pillars in Table IX .
were two inches shorter than that .
This reduction , as well as others of a similar nature , was made on the supposition that the strength of pillars of the same thickness varies inversely as the 17th power of the length , as was found to be the case in solid pillars .
It had been shown , Table VI .
, that a solid cylindrical pillar , with rounded ends , and an enlarged middle , was stronger than a uniform pillar of the same weight .
There seemed little doubt that the same would be the case in hollow pillars , or that a hollow pillarwith an enlarged middle , having the metal everywhere of uniform thick ness , would be stronger than a hollow ouniform pillar of the same weight and thickness of tmetal .
To form these pillars with much regularity in small experimental castings , would have been difficult ; but some idea of the influence of increasing the strength of the middle , might be obtained by reducing the thickness of the uniform pillars , near to the ends , leaving the middle the samrne .
The pillars , in experiments 7 and 8 .
of Table VIII .
, in which the ends were rounded , together with one or two others , were reduced in this manner .
The pillar was reduced to hlalf its thickness of metal , in a band two or three inches broad , near to each end , and to tehree-fourths of its thickness , in a similar band , about half the distance between the middle and each end .
These pillars did not break in the reduced parts ; and from the results it would seem doubtful , whther the strength had been in any degree lessened by the partial reduction of thickness .
If this reduction had been carried uniformly on , decreasing from the ends to the middle of the pillar , leaving it the same there , the pillar would but have had three-fourths of the metal in it , and its strength would , perhaps , have been but little decreased .
Of the first ten pillars in Table IX .
, being those with their ends flat , one half were reduced by having portions cut away from each , as before described .
These pillars never broke in the reduced parts first , except in one instance , and that xvas where the thickness , half-way between the middle and the ends , had been reduced sorewhat more than one-fourth , which was the quantity ordered to be taken away from that part .
The calculated values in Table IX .
, taking a mean from the five reduced pillars , give 3060 ; and from the five which were not reduced , 3308 ; the latter being t13 stronger than the former , leaving it doubtful whether reducing the strength toward the ends , in pillars with flat ends , effects any saving of metal .
The experiments in Table VII .
, uLpon solid pillars , with discs upon the ends , show a saving , however , when the strengtl in the middle is much greater than toward the ends .
The short pillars in Table X. will be referred to further on .
In able XI .
, st two pillars were f thest two pillas were from the same inodel .
They were about 2 inches diameter in the middle , and 1 ' inch near to the ends , which had strong discs upon them turned flat , to give them a firm bearing ; the pillars were hollow , as before .
The weight of the former was 36lbs .
, and of the latter 369 Ibs .
, and their breaking weights 28667 and 27491 , mnean 28079 Ibs .
From a mean between experiments 5 , 6 , 7 , Table IX .
, it appears that a hollow uniform pillar 2 02 inchles external diameter ( the mnean of the above nearly ) , and weighing 361 lbs. , required 30518 lbs. to break it .
This seenis to show that the uniform cylinder is the stlronger* .
STRIENGTlH OF PILLARS .
The third and fourth pillars in Table XI .
, are two hollow pillars 24 inches external diameter at bottom , and 1-3 at top nearly .
They tapered regularly from the bottom to the top , where there was ac disc to increase the stability of fixing .
The mnean weight of these was 36*8 lbs. , and the mean breaking weight 29488 lbs. But we have seen above , that a uniform hollow pillar 2-02 inches diameter , and 36lbs .
weight , would require 30518 Ilbs .
to break it .
Therefore these pillars , like the former ones , are weaker than uniform hollow pillars .
Experiment 5 .
was with a uniform pillar having rounded ends .
The section was of the form +m three inches across ; the pillar weighed 62 lbs. , and was broken with 17578 lbs. But a uniform hollow pillar 3'01 inches diameter , and 501 lbs. weight ( Experiment 13 .
, Table VIII .
) , required 26707 lbs. to break it .
Whence we see , that pillars of this form , which is nearly that of the connecting rod of a steam-engine , are very weak compared with uniform hollow cylindrical ones ; the strength being as the 1'88 power of the weight nearly ( Art .
46 .
) , the hollow cylinder , if of the same weight , 62 lbs. , as the pillar above , would have required 39645 lbs. , or more than twice 17578 , to break it .
Experiment 6 .
is the result from a model differing from the last in having the section of the form H. This was considerably stronger than the preceding , but much weaker than a hollow cylinder of the same weight .
Experiments 7 and 8 .
show the strength of solid pillars when the pressure is applied to some point intetrmediate between the centre and the circumference ; and exhibits a great falling off , from what it would have been , if the pressure had been through the axis , as in pillars from the same model in Table I. Experiment 9 .
is on a pillar so fixed that the pressure is in the direction of the diagonal ; and comparing its strength with that of one of the same dimensions , with rounded ends , in Tab)le I. , it is nearly the same .
The strength is , therefore , only one third of what it would have been , if the pillar had been erect , with its ends well supported , as in 'Table II .
The long pillars referred to in this paper will , I conceive , comprise most of the forms at present used in practice , including the best of those employed by the ancients .
With respect to pillars so short as to break by crushing , experiments on twenty-eight species of timber are given at the end of the paper ; and on this subject I would beg to draw the attention of the reader to my Report on cast iron , in the Sixth Volume of the British Association , before referred to .
The experiments upon crushing , by M. RONDELET ( Art de Batir ) , and by Mr. GEORGE RENNIE , will be known to most of my readers .
pillar , and D ' , E the diameters of the two ends of that in the form of a truncated cone .
But if we compute the diameter of a uniform cylindrical pillar of the same length and solid content as one with unequal diameters , we shall find the uniform pillar stronger than the other , and the more so according as the inequality of the diameters of the latter is greater .
413 Computed Strength of Hollow Cast-Iron Cylindrical Pillars .
49 .
In solid cylinders of the same material , the strength to resist breaking varies according to some constant power n of the diameter ; the length of the pillar and form of the ends being the same .
Thus , strength ox D ' , where D is the diameter .
In cylinders with rounded ends , my experiments give n= 3'76 nearly ; and in those with flat ends , n= 3'55 nearly .
According to the theory of EULER , the strength of a hollow cylinder , to resist incipient flexure , varies as the difference of the fourth powers of the radii , or of the diameters , external and internal , or as D4 d4* , where D and d are the external and internal diameterst .
It appeared probable that this ratio , modified by changing the indices to the results of the experiments upon solid pillars , might agree with those upon hollow ones .
This was tried , and , with great satisfaction , found to answer .
We have , therefore , Strength as D376 d3'76 , in pillars with rounded ends ; Strength as DD " ' d " ' , in pillars with flat ends .
To determine the strength of a solid pillar of a given length and one inch diameter , calling that strength x , and putting w for the breaking weight of a hollow pillar of the same length whose external and internal diameters are D and d , we have D3'76 d3'76 1376 376 w : .
Whence x D3.70d37 for rounded ends ... x D355d ' _ 55 for flat ends .
By these formule , the strength of a solid cylinder , 1 inch diameter and 7 feet 63 inches long , has been calculated from all our longest hollow cylinders , and inserted with the results of the experiments from Tables VIII .
and IX .
The value of x thus obtained from each of the experiments , must , it is evident , be nearly constant , if the assumption respecting the indices be correct .
I have extracted these values from the different experiments and given them in the following Table , and also a fraction indicating the magnitude of the difference between the result from each experiment and the general mean , and consequently the error which would arise from calculating by the general mean .
t-.* 0 +54 2522 1+o1+ 7"7 952 l5'6 15 11+ 567 1 133 13'7 1+ 2.5 2973-70 3031-50 2968-70 3586-90 3573'80 3290-30 3214-80 2988-30 3099'00 3115-50 3207-70 I+ 13.9 +1 19,5 + 1316 1 -'*Q.2 19 ' Il S'9 12 X 15'1 1+ 1-151 1 35*1 1 144 11 4-9'9 00 415 50 .
The last value of x , in pillars with rounded ends , was from a pillar so short as evidently to have been a little crushed by the breaking weight , a matter which we have seen weakens a pillar ( Art .
6 .
) .
If the results from this experiment had been omitted , the general mean would have been increased from 9323 to 941 , and the fractional errors somewhat lessened .
Another cause of irregularity in some of these pillars has been before alluded to ( Art .
47 .
) .
The mean values of x , above , represent the strength of a solid pillar one inch diameter , and 7 feet 6inches long , as obtained by calculation , on the preceding principles , from pillars whose length wasirty tes th e diameter , or upwards , in those with flat ends , and half that length or more , in those with rounded ends .
It was shown ( Art .
31 , 32 .
) , that when the lateral dimensions of pillars are the Pillars with rounded ends .
Pillars with flat ends .
Error from using Error from using Values of .
general mean .
Values of .
general mean .
, w.M.g , erl ... .a .
834-371 933-13 826-20 903-28 808-21 1041-70 917-62 996-24 1123'50 989'95 949-48 1059-50 819.46 895-02 880-11 1026-60 927-86 1005-30 784-90 1r I. same , the strength varies inversely as the 17th pQwer of the length , as a sort of general average .
Adopting this number , and calculating by it the strength of a pillar one foot long , from the strength of one 7 feet 63 inches long , we have the strength of solid pillars one foot long and one inch diameter , In pillars with rounded ends 29074 lbs. = 1.2979 tons ; In pillars with flat ends .
99318 lbs. 44'338 tons .
The mean strengths of pillars one foot long and one inch diameter , as obtained ( Art .
35 to 37 .
) from the experiments on solid pillars , are 33379 lbs. and 98922 lbs. respectively ; which , in pillars with flat ends , is pretty nearly in agreement with that above .
But the defective specimens in the first hollow pillars , with rounded ends , has caused the number to be 29074 , whilst the solid cylinders gave 33379 lbs. 51 .
Taking the numbers above , as deduced from the hollow pillars , we have for the strength of hollow cylindrical pillars in general , D3-76_ 76 29074 117 , in pillars with rounded ends ; D3.55 _ 3'55 99318 1.7 , in pillars with flat ends .
In using these formulae , which answer for solid( cylinders when d= 0 , it must be borne in mind that they do not apply to pillars whose length is less than about thirty times the external diameter when the ends are flat , nor to those of less than half that length when they are rounded .
Short hollow Pillars .
52 .
When pillars are shorter than as above , we have seen ( Art .
6 .
) that there is a falling off in their breaking strength , on account of a change being produced in the material through the great weight necessary to break them .
This falling off has been attempted to be accounted for ( Art .
41 .
) , and a formula given by which the strengths of pillars , however short , could be calculated from the theorems used for long pillars , by means of the crushing strength of the body , the latter being shown to be an element in the resistance of pillars to fracture by flexure .
This formula for short pillars ( Art .
43 .
) is bc Yb+c where b is the strength of the pillar , as calculated by the rules for long pillars ( Art .
51 .
) , and c the weight which would be required to crush the pillar without flexure .
Table X. contains the results of experiments upon thirteen pillars varying in length from twenty-four times the external diameter down to less than eight times .
In this table , the deflections were not observed , as they were very small , and r irquired uch care ; and there was considerable danger in observing them , for the pillar usually broke , with violence , into many pieces .
They were made , however , with great care : 416 the castings were good , and were , as before remarked , frequently bored within and turned on the outside , to ensure uniformity as much as possible .
The values of b , c , and y , are calculated and inserted in the Table .
I will extract from it the real and calculated breaking weights , that the reader may see how far they agree .
Real breaking weight .
Calculated breaking Difference in parts of Real brean weight .
real breaking weight .
33679 32331 _ 24.9 32867 31790 -_ L 35302 36501 +1 31195 28764 30383 30291 -41751 40128 -25 27135 29449 +1 25511 26191 + 375 25105 26273 +2 26729 27364 +4 27135 30863 +7 37285 40257 + -1 34037 31750 --4Power of Cast-iron Pillars to resist long-continued Pressure .
53 .
In all the previous experiments , the pillars were broken without any regard to time , and an experiment seldom lasted longer than from one to three hours .
There might , therefore , be considerable doubt upon the minds of many persons whether the results obtained would be consistent with those which would arise from long-continued pressure .
At my suggestion , therefore , Mr. FAIRBAIRN had the apparatus ( fig. 2 .
Plate XIV .
) erected , by which pillars might be permanently loaded .
Four pillars from the same model were used ; they were each placed vertically upon a horizontal iron plate on the ground , and pressed upon at the top by means of a pin , kept vertical by passing through an iron collar surrounding a hole bored in the horizontal timber at the top of the frame .
The bottom ends of these pins were slightly concave , that they might press , without slipping , upon the top of the pillars , which were rounded at the ends , whilst the top of the pin was rounded , that it might be acted upon uniformly by the square flat plate which rested upon it , and which by means of the suspending rods at the corners supported the weight .
Mr. FAIRBAIRN intends that the results of these experiments shall continue to be observed in conjunction with his interesting experiments upon the effect of long-continued pressure upon horizontal bars .
The results obtained up to the present time are below , as taken from Mr. FAIRBAIRN'S register .
Uniform pillars of Low Moor Iron , No. 3 , each six feet long and one inch diameter , rounded at the ends , and placed vertical with the following weights on each ... ... ... .
1st Experiment .
2nd Exriment .
3rd Experiment .
4th Experiment .
ate of obser .
TemperaWeight of pillar Weight of pillar Weight of pillar Weight of pillar vation .
ture FAHR. : 14lbs .
13 oz. 14lbs .
13 oz. 141bs .
1 oz. 141lbs .
13 oz. Permanent load .
Permanent load Permanent load Permanent load i 4481bs .
7841bs .
1120 lbs. ; 14561bs .
Deflections .
Deflections. .
Deflections .
' Deflections .
1839 .
' inch .
inch .
inch .
July 5 .
72 '018 -03 6 .
68 ii *'215 250 8 .
c 69 '01 '02 230 .
-275 16 .
63 *01 -023 ' 235 '310 23 .
64 '01 '02 '235 *320 Aug. 15 .
63 '01 *023 '235 -395 Nov. 7 .
50 01 ' -03 8 '24 a -825 Dec. 9 .
39 '01 O'025 '243 -955 1840 .
vf Broke about 19th Feb. 14 .
50 '01 '025 , *245 or 20th December , 1839 .
April 27 .
63 '01 : -025 ' 365 June 6 .
i 61 ' -01 '025 *380 Aug. 3 .
~ 74 ' ' 01 -030 - ' 395 Sept. 14 .
55 '01 '025 '380 A pillar of the same dimensions , and cast at the same time as those above , was broken with 1500 lbs. laid gradually upon it .
Hence , supposing the strengths to be 1456 97 the same , we see that the pillar in the fourth experiment above , had borne y5 , aJ ' or 7 of its breaking weight , between five and six months ; though the deflection had been regularly increasing from the first .
I would therefore infer , that the short or long time employed in making an experiment has little influence upon its result .
In concluding my inquiries into the strength of cast-iron pillars , I shall give the tensile , compressive , and transverse strengths , together with the specific gravity of the iron made use of , to serve as data for other researches .
Tensile Strength of Low Moor Iron , No. 3 , or resisWtance which it would offer to a direct force tending to tear it asunder .
54 .
The castings were of the same form as were used for this purpose in my Report on the Strength of Cast Iront .
The form of the section torn asunder was + ; and the force was in a line perpendicular to this , and passing through the centre .
The nds of the castings were made very strong and rigid , and they were perforated , so that links passing through them , to tear the castin asunder by , should act directly through the axis , square ins .
1st exp .
area of section 1*53301 M Breaking weight 212191 Mean 22385 , 2nd exp .
area of section 1'5470 Breaking weight 23551 or strength , per square inch , 14535 lbs. = 6tons nearly .
Compressive Strength of the Iron used above , or that which would crush it without Flexure .
55 .
The first four experiments on this matter were on rectangular masses , cut out of the pieces torn asunder , the sides being rendered smooth and perpendicular to each other .
The other are from the shortest cylinders in Table II .
_No .
of ?
Area of section in inches .
Height of Crushing weight .
Crshing weigh Experiment .
specimen .
~ ' per square inch .
sq .
in .
inches .
lbs. lbs. 1 1-00 x -25 = -25 -50 59 3 ?
0 X25=25 @50 25923}Mean 25923 103692 2 1.1 1,100 x *^5.25.50 295923 Mean ~59v3 3| 1-00 X 25 = 25 1'00 24355 er4J 1-00 x 25 = 25 100 24747 Ji 451 ; 9804 5 o &lt; i*21237 1.00 239631 6 100| -21237 1l00 24747 Mean 24616 115911 7j &gt; 21.37 1.00 25139J 8:.-21237 9'00 35| 9 J. J21237 2.00 22699 f 2287 107675 When bodies are crushed , as above , they give way by a wedge sliding off in an angle dependent upon the nature of the material ; and in cast iron the height of this wedge is about 1| the diameter , or thickness of the base of the wedge .
If the body to be crushed is shorter than would be sufficient to admit a wedge of the full length to slide off , then it would require more than its natural degree of force to crush it ; because the wedge itself must either be crushed , or slide off in a direction of greater difficulty .
If , on the other hand , the height of the body to be crushed be much greater than the length of the wedge , then the body will sustain some degree of flexure , and fracture will be facilitated in consequence .
To determine the crushing strength of the iron above used , I shall , for the reasons just given , take the results from those specimens which are but little longer thatn the wedge .
Taking , therefore , a mean from the results of the rectangles half an inch high , and the cylinders one inch , we have , From the former 1036921 From the former 103692 Mean 109801 lbs. = 49'0183 tons per square inch .
From the latter 115910 Whence we obtain 86238 lbs. = 38*499 tons , for the weight which would crush a3 H2 419 short cylinder one inch in diameter , since the crushing force is as the area of the section , as I have shown* .
Transverse Strength of the same Iron ( Low Moor , No. 3 .
) .
56 .
To determine this , the results from two square bars placed upon horizontal supports , four feet six inches asunder , and broken by weights suspended from the middle , are given below .
First Experiment .
Depth of bar 1'002 inches .
Second Experiment .
Depth of bar 1'005 inches .
Breadth of bar 1'004 inches .
Breadth of bar *996 inches .
Weight .
Deflection .
Deflection .
Deflection .
Load removed .
Load removed .
lbs. inch .
inch .
lbs. inch .
inch .
28 ' 069 -002 ? ?
28 060 56 -145 -006 56 -141 -005 112 -307 -020 112 '321 -030 224 '677 '07 224 -725 336 1-143 '17 336 1-230 392 1-413 '26 392 1'550 448 1-742 '38 448 1'910 461 broke .
473 broke .
riiiiiiiiii li ll The weights were increased toward the last by about 7 lbs. at a time .
The modulus of elasticity in lbs. per square inch , obtained from the deflection caused by 112 lbs. is , From the first experiment 14,251,950 13,918,740 Mean .
From the second experiment 13,585,530 Specific Gravity of the same Iron .
57 .
This , from the results of six experiments , is as below : 7122 from cylinders three-fourths of an inch diameter .
7135 from cylinders three-fourths of an inch diameter .
7034 from cylinders three-fourths of an inch diameter .
7090 from cylinders half an inch diameter .
6977 from cylinders half an inch diameter .
6952 from a hollow cylinder .
Mean 7052 .
Pillars of Wrought Iron , Steel , and Timber .
58 .
The experiments upon these , which are given in Tables XII .
and XIII .
, are similar in their objects to those upon cast iron , but muchl more limited in their extent .
They were made , indeed , principally to confirm views which presented themselves during the experiments upon cast-iron pillars .
The following abstract contains the lengths , lateral weights of the three kinds of pillars tried .
dimensions , and breaking Pillars with both ends Pillars with one end flat , and Pillars with both ends flat .
rounded .
the other rounded .
Length .
Diameter .
Breaking weight .
Diameter .
Breaking weight .
Diameter .
Breaking weight .
inches .
inch .
lbs. inch .
lbs. inch .
lbs. g 903 1-017 1808 1.02 3355 1'02 5280:7 60- } 1'015 3938 1-03 8137 1'02 12990 i , 30k 1.015 15480 1-015 21355 1.015 23371 E 301 1.015 15480 1.015 21.187 disc .
1-015 25387disc .
15~ 1.005 23535 1.005 26227 1.005 27099 | 29-95 -87 10516 '87 20315 87 26059 Side of Side of Side of.~ square .
square .
square .
60 ?
1-75 3197 1.75 6109 1.75 9625 The pillars in each horizontal line in this abstract are of equal length , and nearly of equal diameter , the only essential difference being in the form of their ends .
Taking the longest pillars , those of 903 and 60* long , the strengths are Ends rounded .
Ends rounded and flat .
Ends flat .
In iron ... 1808 3355 5280 In iron ... 3938 8137 12990 In timber. .
3197 6109 9625 The strengths in these three cases are , therefore , as one , two , three nearly , the middle being an arithmetical mean between the other two .
These results are a corroboration of those from cast-iron pillars ( Art .
16 .
) .
We see , too , from the wroughtiron pillars 304and 15w inches long , as well as from those of steel , which appeal to have been slightly crushed by the weight ( Art .
6 .
) , that this ratio one , two , three , does not obtain in short pillars , but that the difference of the strengths of the three forms of pillars becomes less , according as the number of times which the length of the pillar exceeds the diameter decreases .
But whatever may be the ratio of the strengths of pillars with rounded ends to those with flat ones , the strength of those with one end rounded and one flat , is nearly an arithmetical mean between the strengths of the other two , as will be seen from the numbers below .
Mean from strengths of Strengths of pillars , one pillars , with both ends end rounded and one rounded and both flat .
flat .
lbs. lbs. 3544 3355 8464 8137 19426 21355 20434 21187 25317 26227 18287 20315 6411 6109 421 We have seen in the preceding abstract from Table XII .
, that the longest wrought-iron pillars of the three classes there considered , have their strengths as one , two , three nearly ; but that the shorter pillars , those of 30 : inches long , or less ( the diameter being 1'015 inch nearly ) , differ much less in their relative strengths .
These in the 30inch pillars being Ends round .
Ends round and flat .
Ends flat .
15480 21355 23371 15480 21187 25387 The second and third of these numtnbers in either series are much less than twice and three times the first number , which was a mean from three experiments .
We see , therefore , that these second and third numbers are the results from pillars which were affected by some cause that did not reduce the strength of the pillars of equal diameter and length , which were broken by the first weight , 15480 lbs. For it will be seen further on , that this first weight was as great as is in accordance with the results from the longer pillars .
60 .
To determine the cause of this anomaly , which had first been exhibited in the experiments upon cast iron , and afterwards was found to take place in wrought iron , steel , and timber , I felt to be very difficult .
The experiments upon cast iron offered no satisfactory explanation of the matter , for it usually changes but little in form before fracture .
Wrought iron appeared , therefore , to be a better material for exhibiting the cause .
Three small cylinders of the same iron as that in Table XII .
were therefore formed .
They were each exactly 2i inches long and T62 inch diameter , their ends being turned flat and perpendicular to the axis ; they were subjected to compression with their ends well-bedded , and after bearing certain weights they were tried by a gauge which had exactly fitted them before .
The results are in the table below .
No. of Weight Weight per square Reduction in length .
marks Experiment .
laid on .
inch .
lbs. [ lbs. tons .
inch .
1 6222 ~ 20610 = 9'2 0 ?
No alteration , or , if any , extremely small .
10 7342 24319 = 10'9 '015 1v 8462 28028 = 12*5 *02 Diameter sensiblyincreased , cylinder not bent .
2s 12444 '* 41218 = 18-4 Diameter increased from -62 to '63 inch , 2 12444 41218=1814 ' 04 3 18667 61830 = 27'6 *16 Bent a little , diameter increased to *65 inch .
We see , from the above , that 6222 lbs. produced no sensible change in the cylinder *62 inch diameter , but that 7342 lbs. and 8462 , particularly the latter , which was equal to 12i tons per square inch , produced very obvious changes .
Since the resistance to crushing are as the areas , the former of these two values would give 19677 lbs. , and the latter 22679 lbs. , for the pressures which would produce an equal alteration in a bolt 1'015 inch diameter ; which is the diameter of the 422 bars upon which most of the experiments on wrought iron were made .
But we have seen ( Art .
59 .
) that 21187 lbs. , and 21355 lbs. produced , a great change in the strength of the pillars 301 inchles long ; it was , therefore , obvious that incipient crushing was the cause of the great falling off in the strengths of these pillars , and indeed of those in all the preceding experiments ( Art .
6 .
) ; and a knowledge of this fact prepared me for attempting to adapt the formula used for long pillars to the case of short ones ( Art .
41 .
) .
61 .
It was shown ( Art .
12 .
) from the experiments upon cast iron , that a uniform pillar , with its ends flat , had the same strength as one of equal lateral dimensions , and half the length , with the ends rounded .
Making a similar comparison between the wrought-irol pillars , in Table XII .
, or in the preceding abstract ( Art .
58 .
) , we obtain an approximation to the same result ; and that even when the pillars , from their shortness , are much crushed by the breaking weight , which we have seen was the case with the shortest wrought-iron pillars .
Strength of Pillars , of Wrought Iron and Timber , as dependent upon their dimensions .
62 .
On this subject I have laboured under considerable difficulty , for want of experiments of a sufficiently comprehensive character to furnish all the data that were required .
The complete determination of the constants for the purpose would require experiments upon pillars , whose lengths , compared with their diameters , were nearly as various as those upon cast iron in this paper .
I trust , however , that the experiments in Tables XII .
and XIII .
, and the analogies derived from the results on cast iron , will supply the deficiencies so far , as , in some measure , to meet the wants of practice .
Length.-To find the inverse power of the length to which the strength of wroughtiron cylindrical pillars , of a constant diameter , is proportional , I shall follow the mode used for cast-iron pillars ( Art .
26 .
27 .
) ; and confining the inquiry to pillars sufficiently long not to have been crushed ( Art .
6 .
60 .
) with the breaking weight , we have as below : Lengths of which the strengths : D Lengths of which tb srengths Description of pillar .
Inverse power of the length .
ft. ins .
ft. ins .
7626 Rounded ends .
19546 7 6 .
5 0O| Rounded ends .
19199 Mean 2'0316 .
7 6 .
5 0| Flat ends .
22203J Whence it appears that the strength is as the inverse square of the length nearly .
Diameter.-To find the power of the diameter , or of the side of a square , to which the strength of wrought-iron pillars of a given length is proportional .
For this purpose we have , from Table XII .
, the following results : In the pillars 2 feet 6Q inches long , rounded at the ends , from a mnean between three experiments , one of *52 inch diameter had its greatest resistance overcome with 423 Ibs ; and from a mnean between three other experiments , a pillar of the same length and 1P015 inches diameter had its greatest resistance overcomrne by 15480 lbs. Since 1O = 91 9 , we have , putting n for the power , 1 : ?
19519 : : 1260:15480 .
Whence reducing , and taking the logarithms , we obtain n=3 75073 , which is nearly the same as the mean from the cast-iron pillars with rounded ends ( Art .
23 .
) .
Computed Strength of YWrought-iron Pillars long enough not to be crushed , as in Art .
59 .
and 60 .
, by the weight which would overcome their greatest resistance .
63 .
We have seen , from the last article , that the strength of cylindrical pillars whereof the length is constant , varies nearly according to the same power of the diameter as that in cast-iron ones .
But where the diameter is constant , the strength is inversely as the square of the length , nearly .
Whence if , in round or square pillars , we put d for the diameter , or side of the square , I for the length , b the breaking weight , a , a ' constant quantities , and adopt the powers of d , which were used for cast iron , we shall have ( Arts .
36 .
38 .
) a d3'76 b--12 ~ , in pillars with rounded ends ; a ' d3,55 b12 in pillars with flat ends .
From the pillars in Table XII .
we obtain as below : ft. ins .
lbs. From those 7 6| long , with rounded ends , a 97049 From those 5 0 ?
long , with rounded ends , a= 94648 Mean value of a= 95848 ft. ins .
lbs. From those 7 6long , with flat ends , a'= 281464 From those 5 0 ?
long , with flat ends , a ' = 307770 Mean value of a ' = 299617 Here the length is taken in feet , and the diameter in inches .
Timber .
64 .
To find n the power of the diameter , or of the side of a square , to which the strength is proportional , the length being constant , and the pillar so long as not to be crushed ( Arts .
59 .
60 .
) with the breaking weight .
For this purpose Table XIII .
will supply us with the results of six experiments upon pillars of Dantzic oak , 46 1 inches long .
Whence it appears that a pillar 1P02 inch square was broken with a mean weight = 1754 lbs. , and one 1'50 inch square with 7888 lbs. 424 Herej.j2= 7 , .
' .
n : 1-47f : : 1754 : 7888 .
Reducing gives n= 3'902 .
Whence the strength is nearly as the fourth power of the side of the square .
I made no experiments to ascertain the inverse power of the length to which the strength is proportional in timber .
The great flexibility of such long and slender pieces as would have beeii necessary to give sufficient variety in their length , and for the length of the shortest to be a sufficient number of times the thickness , that it would not be sensibly crushed by the breaking weight , -convinced me that it would be as well to seek for this power hypothetically , as to endeavour to obtain it by experiments , which would probably be unsatisfactory .
I therefore assumed ( after trying some other hypotheses less successfully , ) the strength of timber to be as -2 , which is the simple conclusion of EULER .
ad4 Whence b= -2 , where the constant a must be obtained from experiment , for any particular kind of wood .
Dantzic Oak .
The experiments in Table XIII .
upon square pillars of this wood flat at the ends , give as below : 1st .
From a mean of the results of four experiments , a pillar 60 inches= 5'04166 feet long , and 1-75 inches square , required 9625 lbs. to break it .
Whence , taking the length in feet and the lateral dimensions in inches , we have ( in the for , ad4 b 12 mula b= g- ) b-= 9625 , d= 175 , 1= 5'04166. .
a= 4= 26085 .
2nd .
From a mean of three experiments upon pillars 46-1 inches = 3,84166 feet long , and 1X50 inches square , the breaking weight was 7888 lbs. ; which gives a , obtained as above , = 22999 lbs. d4 A mean between these two values of a is 24542 lbs. ; whence b= 24542 dHere the length is in feet , the side of the square in inches , and the weight in lbs. ; and the rule is only applicable to cases where the pillar is in length so many times the thickness that it is not crushed , as in Articles 6 .
and 59 , by the breaking weight .
These remarks apply to the following cases : Red Deal .
The results from two experiments upon square pillars of this wood , each four feet d4 ten inches long , and two inches square , Table XIII .
, give b= 17511 , ; the different values being as above .
Experiments upon Oak ( from TREDGOLD'S Carpentry ) .
These experiments were upon square pillars of different lengths , varying as 1 , 2 , 3 and the sides of their squares as 2 , 3 , 4 .
This writer has , in most instances , given four experiments or more to determine the breaking weights of pillars of each size .
The longest pillars , compared with their thickness , in this series are 6*375 feet long , and 24126 inches square , the length being about thirty-five times the side of the square ; and the average breaking weight from four experiments was 7769 lbs. We therefore get a d415455 lbs. , whence b= 15455 -72 Further on , I shall endeavour to compute the results of LAMANDE'S experiments .
Short Pillars of Timber .
65 .
We have seen that pillars of cast iron , wrought iron , and timber , follow in a great measure the same laws as to their strength ( Arts .
5 .
58 .
) .
In these different materials , pillars with rounded ends have their strength in a constant ratio to that of those with flat ends .
This , however , only applies to pillars whose length is so many times the diameter or thickness , that they are not sensibly crushed by the breaking weight .
My experiments have not shown what part of their ultimate crushing weight pillars of timber will bear without suffering such a diminution of their strength , from this cause , as to need correcting for , as was done in estimating the strength of short pillars of cast iron ( Art .
41 .
) .
In that metal there appeared to be a slight falling off in the strength of even the longest pillars , which I accounted for by the shifting of the position of thie neutral line , but it was not such that the results of the general formulae then used ( Art .
36 to 38 .
) needed modifying ; nor was there shown a defect in comparing the breaking weight of the pillars with rounded ends with those with flat ones , till the breaking weight was about one-fourth of the weight which would have crushed the whole pillar without flexure ( Art .
6 .
) .
It seems probable , from the experiments in Table XIII .
, that timber will bear , without much reduction of the breaking weight from crushing , considerably more than one-fourth of the ultimate crushing weight ; for the pillars 1'75 inch square and 60 inches long , with flat ends , exhibited no reduction in comparison of strength with those with rounded ends , though the mean breaking weight of the former was more than one-third of the weight which would have crushed the whole section .
It may , however , be remarked , 'that the circumstance of a pillar with flat ends bearing upwards of three times as much as one of the same diameter with rounded ends , does not prove that the former had not its breaking weight in some degree reduced by crushing , since both descriptions of pillars were considerably compressed .
In the absence of other information , therefore , I shall in calculating the strengths of short be pillars of timber by the formula y= 6-b ( Art .
42 .
) , suppose d , as in cast iron , pilas f imerbyth frml yb+cd 426 to be one-fourth of c , the crushing weight of the section .
We have , therefore , as before , bc Yb +3c where b= the computed value from the general formulae for long pillars of different kinds of woods in the last article , and y that for short ones .
If there were no fixed pressure at which pillars in breaking suffered a marked diminution of their strength , but that there was a regular falling off with all weights from the least necessary to break a long pillar up to such as would crush it , then d might be taken = 0 , and the formula would be bc YB+ -c ' where b would have to be obtained from experiments upon the longest pillars only , and this formula must be used for all others but them .
We see that the strength of short ( and perhaps long ) pillars to resist fracture by flexure depends upon their resistance to crushing .
Before attempting , therefore , to calculate the strengths of the short pillars broken by LAMANDE , we must know the crushing strength of the French oak , which fortunately is given by RONDELET* , who states that its strength , from a mean between his experiments , is 6336 lbs. per square inch .
Whence we are enabled to obtain an approximation to the value of c , the crushing weight of the pillars which LAMANDE made use of .
66 .
The following table contains the results of his experiments upon short pillars cominputed by the formula for y , from the value of b , deduced ( Art .
64 .
) fromin his longest and slenderest pillars , which we may consider not to have been materially reduced in their strength by crushing , and which are placed the first in the Table .
months longer .
The lengths of the latter specimens were , in some instances , only one inch , which reduction would increase the strength a little .
But the great difference frequently seen in the strength , as given by the two columns , shows strongly the effect of drying upon wood , and the great weakness of wet timber , it not having half the strength of dry .
The strength of short columns of the same material I have shown to be directly as the area of the section* .
This seems easy to conceive , as bodies of the same nature always become crushed by sliding off in an angle , which is nearly constant ; the height of the wedge , which would slide off , being in timber usually about half the diameter , or thickness of the specimen .
Relative Strength of Long Pillars of Cast Iron , Wrought Iron , Steel , and 7Timber .
68 .
To compare the strengths of pillars , of the same size , in the different materials used in the preceding experiments , I have computed , by the rules used for cast iron ( Art .
36-38 .
) , the strength of a pillar , of the same diameter and length , in each body , taking the results from the longest pillars , in order that they may not be crushed ( as described in Art .
6 .
) by the breaking weight .
The relative strengths are as below : Calling the strength of cast iron 1000 , we have for wrought iron 1745 , cast steel 2518 , Dantzic oak 108'8 , red deal 78'5 .
These numbers , except the last , are all computed from the pillars with rounded ends .
( Continued .
) ent Mean:13Deflection CorreBreaking Mean from ; Length .
Diameter .
Diameter .
of middle sponding weight .
breaking Remarks .
of pillar .
weight .
weights .
inches .
inch .
inch .
in ch .
lbs. lbs. lbs. 20-1666 '785 717 2 ... ... ... ... ... .
55 1r eight 2 lbs. 81 oz. 20-1666 *75 7'7 ... ... ... ... .
5950 J 6602 Weight 2 lbs. 6oz .
15-125 *50. .
-05 1689 Er *)'09 1801 ti '* 111 1913 fr Neutral line well defined .
*31 1997 1997 T:C : : 39 : 11 .
15'125 50 ... ... . .
-20 1801 fT : C::39:11 .
'6 ' *34 1s857 1857 q In this experiment the deflectionwas taken 15-125 ' *50 ... -08 I 1353 t as the pillar was bending .
15 1689 '50 '20 1801 1857 1904 15-125 -77 10138 ... ... . .
T:C : 48 : 29 .
15'125 76 .
9746 ... ... . .
T : C : : 53:23 .
15'125 '75 '76 ... ... ... ... ... .
7786 9223 T : C : 49 : 26 .
For the forms of fracture , see g. Plate XV .
15-125 -99 ... ... ... ... ... ... ... ... 20163 ... ... ... T:C : : 57:42 .
15-125 '99 ... ... ... ... . .
19239 1 ... ... .
T:C::60:39 .
15'125 -99 '99 ... 19855 19752 T : C : 60 : 39 .
For forms of fracture , see i , j , k. Plate XV .
10-083 -76 ] ... 16683 1 10-083 ' 76 '*76 , .
16683 17506 The two first failed by the ends becoming 1 0-083 ; '*77 J -- ... ... , j19102 L spht , by a conical wedge forming at them .
10-083 l 7 ... ... ... ... ... .
19152 7-5625 -51 ... ... ..6188..T C::31 : 20 .
7-5625 49 4578 ... ... ... C..::32:17 .
7'5625 -49 -497 " ... ... . .
, , , e. 5019 5262 T : C : : 34:15 .
7*565 '49 1 ^..89 ... ... . .
T : C : : 32 : 42 .
7'5625 77 23893 ... . .
T : C : : 35 : 42 .
7'5625 '77 2J .
22003 22948 T : C : : 41:36 .
' ... ... ... ... ... These pillars were split at both ends .
For frac !
'\ !
' ; tures , see 1 , m. Plate XV . !
i'( fIn these two experiments the area com3'7812.50. .
I5 pressed seemed to have been greater than 3-7812 '50 50O ... . .
e ... ... ... ... ..a 15233\ 15 107.l the extended area .
The ends were split 3.7812 .
*50 Pig .
1 .
Plate XIII .
represents the lever and other apparatus used for breaking the pillars : it has been described in the commencement of this paper ( Art .
2 .
) .
Fig. 2 .
Plate XIV .
shows the manner in which the pillars were subjected to a constant strain , to try the effect of time upon them ( Art .
53 .
) .
Figs. a , b , c , &amp;c. to n , Plate XV .
are the forms of the fractures from some of the solid pillars with rounded ends in Table I. , the pillars as they appeared after fracture being drawn and placed over the forms of the fractures .
Figs. o , p , q , r , s , t , u , v , Plate XV .
are the forms of the fractures from the shorter pillars with flat ends in Trable II .
; and the figures over them represent the appearances after fracture of the pillars they were obtained from , with marks upon the pillars , showing the places where they broke .
All these sections are referred to in Tables I. and II .
In many of these pillars there was a crack after fracture , showing the position of the neutral line ; and in some , the compressed part broke off as a wedge , the form of which may be seen from the marks upon the pillars .
The groups of pillars , designated as No. 1 , No. 2 , No. 3 , &amp;c. Plate XV .
are intended to represent the forms of the different cast-iron pillars broken , with marks upon them , showing where they usually broke .
Those in No. 1 , which may be considered as including the longer pillars in Tables I. , II .
, V. , represent three pillars of the sane material , length and diameter , whose relative strengths are as 1 , 2 , 3 , nearly .
No. 2 .
are pillars differing from those of No. 1 , in having discs upon the flat ends , to give them a larger bearing ; the results showing that a small increase of strength is obtained by that addition .
No. 4 .
represents the forms of the pillars in Tables VI .
and VII .
, which generally gave strengths about one-seventh or one-eighth above those obtained from uniform solid pillars of the same length and weight .
The marks upon them show where they usually broke .
No. 5 .
includes pillars formed to show the effect of defective fixing ; a pillar so placed that the pressure would pass through the diagonal , bearing only about onethird of what it would have done if the ends had been pressed upon through their whole surface ( Table XI .
) .
No. 6 .
The second pillar in this group is intended , by its section in the middle , to represent the connecting rod of a steam-engine and experiment shows that it is very weak compared with a hollow cylinder of the same length , weight , and lateral dimtensions ( Table XI .
, Art .
48 .
) .
The first forIn of pillar in No. 6 .
gave strengths greater than the second , but still less than a hollow cylinder would .
No. 7 .
shows the relative diameters of the hollow pillars in Table VIII .
, broken with rounded caps upon the ends , the caps being shown near to the pillars .
No. 8 .
represents the hollow pillars in Table IX .
, broken , with their ends flat ; the rings round the three first pillars given , show the manner in which some of these 455 were reduced , one half of the thickness of the metal being taken away from the pillar near to the ends , and one-fourth , half way between the middle and the ends .
When no more metal than this was taken away , the pillars were never broken in the reduced parts .
The fourth pillar given in this group was a very good casting , and the section is represented by W below it .
The small piece where the crack is seen , broke out , showing very distinctly the position of the neutral line .
The pillars No. 9 .
are the first four in Table XI .
They were hollow , and the metal in them was cast to be all of equal thickness , except that in the discs .
The marks upon them will show the places in which they broke .
These pillars were somewhat weaker than uniform hollow cylindrical ones of the same length , thickrfess and quantity of metal ( Art .
48 .
) .
The two last pillars in No. 3 .
were of oak ( Table XIII .
) ; and the second of these is intended to show how timber , when much compressed , becomes wrinkled near the ends .
Inis Plate th e pillars are not drawn of their real lengths , but many of them art of half the length compared with the diameter .

