]\gt ; Prof. B. Threlfall and Mr. J. F. Adair .
On the On the yelocity of Transmission through Sea-water of Disturbances of large Amplitude caused by Explosions By RICHARD HRELFALL , , Professor of Physics , University of Sydney , and JOHN FREDERICK ADAIR , Demonstrator of Physics in the ersity of Sydney .
Commumcated by Professor J. J. ISON , F.R.S. Received March 14 , 1889 .
Read March 21 , 1889 .
PART XPEPIMENTAL .
During the latter months of the year 1886 and the first half of the wear 1887 , experiments were undertaken by one of us in conjunc tion with Lieutenant Ottley , R.N. , of Her ship " " Nelson then stationed at Sydney , with the view of obtaining definite information as to the existence or non-existence of the supposed phenomenon of " " explosion by influence It had been lsly cliscovered relfall , Phil. March , that in the of small exploions of minate of mercury under water , there was a of dirccted effect due to want of symmetry in the disposition of obstacles nding the centre of explosion .
This suggested that a possible explanation of the recorded cases of explosion by influence might be found by a careful examination of initial conditions .
Since in or ( linary torpedo work in the initial conditions of osion al.egl'eatly varied , cases of thetic explosion are generalIy believed to have occurred , it was considered that the most hopeful of proceeding would be by endeavouring to obtain sympathetic losions between masses of explosive , fired under such conditions to preclude the possibility of directed action .
To this end small arges of guncotton were suspended in free air by strings , the distance between the charges was sured , and one of them was ( tctonated .
Out of about fifty experiments of this kind we did not stlcceed in obtaining any explosion by influence , the target charge was generally blown to atoms .
A similar series of experients , in which " " 9-oz.\ldquo ; disks of guncotton were wrapped in oil-silk exploded under water , resulted in the same conclusion .
sion of the target charge was only brought about when so close to the exploding charge as to be exposed to explosive violence amply suflicient to account for the explosion .
In some cases the conditions by placing detonators in the target charge , but no excepto the result was noticed .
At this time we in aining a copy of Major-General Abbotl 's 'SubmarIne Mincs and Explosives , ' and there found that ected action had been noticed by him in the case of exploslve mixbures in a case an Velocity of Transmission of Disturbances through .
497 at one end ( p. 66 ) , and in the case of dynamite ( p. 82 ) in a " " tin cnse , with saw-dust tamping\ldquo ; Appendix .
The sympathetic explosion of dynamite and gnlTcotton had also been investigated by General Abbott , with the general resuIt ( pp. 122 et seq. ) that as a distinct mode of explosion it does not exist .
As this concInsion agrees entirely with the one to which we were led , both with explosions by guncotton in ] .
and water , and explosions of guncotton , gelatine-dynamite , mite No. 1 , and diazobenzene nitrate on one side of a iron plate , the target being on the other side , we did not ptlrsue the matter further .
We may remark that we found diazobenzene very good substaucs to experiment on this direction , as it is exceedingly sensitive , but llot so violent as silver fulminate .
Like the latter and iodide of , we believe that its explosion can be caused by the fric.tion of the small crystals of which it is composed on one another .
If two heaps of this substance , placed on a thin iron plate , were protected by cardboard cylinders without ends , resting on the plate supporting the explosive and surrounding the latter , explosion alwftys took place at a gx.eater distance when the supporting plate was sheet iron than when it was a rather thickcr pine board .
to the thinnes , of the iron plate , the amplitude of the vibrations into which ] it was thrown reater than in the case of , and consequently the heap was more thrown about , and exploded ( other being eqnal ) at a reater d. We have to tbank Dr. Helms , Demonstrator in Chemistry at the University , for his kindness in epnring the diazobeuzene nitrate for us ; it is both tiresome to make and delicate to handle .
On of his ' Mines and xplosives , ' General Abbott gives in an appendix an account of a remal.kable explosio1l of 5 lb. of dynamite .
In this experiment axis of maximum effect was bent of the vertical ( i.e. , .
the normal to the rough the shot ) to such an extent as to call comment .
A theory had been proposed by one of us which we account for this abnormal , if it could ) shown it was only an extreme case of a generally occurring enomenol ] .
With the view of tcsting this theol.y , we decided at the best way would bc to find the of ssion of ough , va ying the quantity an nature of the explof , iv and its case , and notin nny irrcgularity that m occur .
' work , have eeded in doing , extendiug tions to gLlllcotbon Nobel 's No. 1 dynamite , and var the quantity of explosive oz. up to .
charges .
The tnnce o the vclocity was measured was varied three times , ) , the least , to 200 greatest .
ho csults h beeu such to justify the of 498 Prof. R. Threlfall and .
J. F. Adair .
On the The Experimenis .
The Site and General ) } ' the kindness of the military authorities we were to use the torpedo station at Berry 's Bay , in the arbour of Port Jackson , as the base operations .
This bay had the advantage of deep water cIose to the shore , of being comparatively quiet , and within about an hour 's journey of the University .
The Sydney lsIarine Board with great conrtesy allowed us to use their steamlaullches for the purpose of getting to the ground , it being out of course of the ferry-boats .
The general plan of the water and coast-line will be understood from an examination of ( hart ( Plate 3 ) kindly prepared us by Mr. R. J. A. Roberts , who also underlook the sir veying of the distance between the observing stations .
We desire to express our indebtedness to Mr. Roberts , for his very accurate determinations to be referred to hereafter .
The general disposition of the arrangements was as follows : piles , each about 45 feet long , wers dri into the mud of the sea-bottom at stated distance , and about 40 yards from the .
To these piles were attached the apparatus recording the ival of due to the explosions ; the connexions the shore made } ) means two lengths of armoured cable , ] lent to us by the officer commanding the torpedo corps .
At first we tried to use erhead wires supported on telegraph poles , and running from the piles to the shore ; these , however , were so systematically carried by passing steamers , that they had to be abandoned .
In a portable constructor 's house on the wharf we erected the pendulum chronograph , to be described shortly , and there established the arrangements for firing the charges and working the dlng apparatus .
About a year was spent in with different forms of apparatus for recording the arrival of the disturbance .
Three forms of " " gauge\ldquo ; were devised , tested , and oned before the ultimate form was hit upon .
Considerable difficulty was also experienced in avoiding missfires , as we desired to avoid complicating by using heavy cases for our torpedoes .
At one time work was entirel .
pended for months , as we could not et any , and were not allowed to store the dynamite which we could buy the market ; guncotton ot be procured here except through courtesy of the Governmen .
The remark applies to torfi , which we were forced devise a mealls of manufacturiug for ourselves .
Finally , owing to the good offices of Major Penrose , R.E. , we obtained a supply of 100 lb. of guncotbon from the Government , and several boxes of detonators , for which we desire to record ollr thanks .
In our opinion any work of the kind should not in the future be underVelocity of ransmission of Disturbances through .
499 aken except such military officers as are in a position to avoid the imposition of difficuIties about the of explosives .
The general course of conducting an experiment was as follows:\mdash ; The gauges were got out , adjusted to their maximum sensitiveness , and generally overhauled .
The whole number of torpedoes for the day 's were and pri med .
The gauges were -en out in a dinghy , and ad.usted to supports on the by one observer .
The electrical cuits were completed and tested , and the .
out drums , and the ends buoyed at yards outside each pile , and in the direct line of the piles .
The temperatore of the water was taken at the , and at a depth of 6 feet .
Meanwhile the other got the adjusted , noted the temperature of the tuning.fork , assul.ed himself of the perfection of the electrical firing ements , and smoked the glass plates .
All being ready , one obserYer ( J. F. A. ) generally assisted by Mr. Proctor , the storekeeper , shipped a Tedo , and proceeded to the f point , and connected ib with the firing wire in the usual mannel The other observel ( R. T. ) mnde fin arrangements in the hut , and ibat all precautions to pl.event premature explosion were ( a to the effecb that he tol.pedo was , he replied by a to the shob and end of the cable , previously efully insul was then connected to ; wire passing inbo the hut .
The plates of the firing battery then lowered into the liquid , and a safet ] was placed in the firing position .
At a signal party in the boat , the pendulum was to fall , ) torpedo , and ' the recording plate ainst the connected with the recording apparatus and the scriber of tlJe tuning fork .
Immediately after firing , the trigger reset , the ) placed in its second position , the battel.y plates lited out of the solution .
and the cable insulated .
It will be seen that no than four separate and independent acts of )ution were thus performed shot .
We have had no accidents of tlJe n of explosions after several huItdred shots , though once in a dirty guncotton disk with a brass , a partial explosion , ppily without any , took place .
After of the shot the in the boat proceeded to row past the other pile , and there fastened the nexb shot to its appropriato buoyed cable .
Meanwhile the erver on shore took smoked plate out of the chronograph , wrote on it whatever data were necessary , allowed the pendulum to swing back , eadjusted the 500 Prof. B. Threlfall and Mr. J. F. Adair .
On the and -fork , replaced the smoked plate , and waited for the signal as before .
The operations then repeated till half the prescribed shots had been { ired ; the were then reversed on the piles , and the same process again gone through .
The greatest number of records ever succeeded in in a day was sixteen .
The greater part of the time was taken up in adjusting the gauges and preparing the charges .
At the end of the work the were brought in , the firirg wires would up , and the plates packed in a box for urement at the University .
Four ] records were obtained on each plate .
Shots were fired alternately from left to , and from right to left , past the piles and gauges , with a view to the elimination of the ants of the recording We have ofol the following experimental matters to explain :\mdash ; 1st .
The measurement of the time of passage of the shock through water , from one gauge to the er .
2nd .
The measnrement of the between the requisite to ensure symmetry of explosion , and of explosion in the line joining the two We shall consider the time-measurements first .
In .
on a chronograph we had to bear in mind it was an condition of success to get an instrnment which allowed of rapid and easy manipulation .
Since we to make absolute measuremenk , that is , to be able to rerluce our indications to mean solar seconds , it follows that some , standard time measurer must be adopted , even at the of othel.wiseleducing the handiness of the and of runn risk of a diminished sensitive .
It is a that the hods which give the most absolute values are seIclom as as which allow of a slight of in the absolute value of their indications .
Considerations such these led to the preference being to an instrument of the pendulum or raph d ption .
Such an instrument WflS therefore commeItced in JanuarJ7 , 1887 , by Mr. Cook , the sity Assistant in Ph ysical Laboratory , ] at the time his appliances were of the most meagre description .
The pendulum was about a yard Jong , bob consisting of a { lat oval mass of lead .
This the supporting and adjusting for the glass plate .
There holdfast catches for the bob at end of its swing .
On the basc of the instrument there were hree s ] ) liances .
stand to support the tuning-fork , and of its adjustn1cnt a plane , and in a plane ttrallel to the of motion of the .
The plane of the ] tter was invariable , as it on hard steel , supported on grooves cut in steel } Transmission of rbances throngh Sea-water .
501 The -fork\mdash ; one of Koenig's\mdash ; making , or rather marked as giving , } vibrations per secornl , was maint electrically by a second scribing apparatus , consisting of two scribers and two electromagnets , could be adjusted through a cousiderable range in a plane , and in a plane perpendicular 10 the motion of the pelldulum ; and also to some extent in a plane parallel to the plane of motion of the pendulum .
There was a good deal of trouble about this part of the apparatus , chiefly from the desire we had lose no sen .
through slowness of action .
To this end the levers of the ribers were made of aluminium , and the nets and spl'ings were very powerful as compared with the moment of of the sc around their axes .
We altered apparatus , almost remaking it several times , as we found that the limit to the accuracy of our time observations in the discrimination the precise point at which the line , traced by the scriber at rest on the smoked glass , began to be } errupted by the motion of the scriber .
The sharper the bend the more the discrimination .
ally we got the scribers to the line with almost complete suddenness , i.e. , at something like to its ginal direction .
The for the was of a kind that will easily occur to the reader .
The catch on the pendulum could be slightly set wards or backwards , and only affected the trigger when the pendulum was moving in direction .
The trigger itseIf could be moved along on arc below the pendulum , so as to allow the firing to take place at such a time as would give the ' the benefit the pendulum 's most motion .
In order to ensure a instantaneous contact , the on being let oif drove the influence of a ) powerful spring ) a platinum-coated wedge between the faces of the brass plngs .
The firing battery consisted of six good bichromate elements with ; and the battery for the ecording gear consisted of twenty Leclanche cells , ten with large plates and ten with small .
In all cases use was made of a good " " return this being secnred permanently at one end , bit of armoured cable , at the piles with plates , and at the torpedo with a bit of old copper wire or thin plate .
rinq Apparntus .
\mdash ; ' ' Gauges This } ) of the experimental arrangement was the most difficult to get to .
At first a of small pendulum was ported so as to han near an india-rubber disk , inside a wide brass cylinder ; the whole was sunk under water , leaving the contact wires protruding .
The cultics , , were such as to end this form of useless .
The form adopted for small explosions , mere noises , in fact , is in 502 Prof. R. Threlfall and Mr. J. F. Adair .
On the trumpet ends in a small tube , closed with an india-rubber film diaphragm .
Against this diaphragm rests an independently supported strip of aluminium foil , as fine and light as possible ; at its end it carries a tiny point of platinum .
The fine diaphragm , with its tube and aluminium contact , is protected an rangement of a glass cylinder , stout brass ends .
Tbrough brass end passes a long brass screw , carrying a platinum disk at its lowest .
This disk is insulated from the brass portion by means of ebonite , and is connected by a wire with an sulated binding outside the apparatus .
screw is adjusted so that it just clears the platinum point 011 the aluminium spring .
the diaphragm is disturbed the point is thrown ainst the disk and makes contact .
In order that the contact may be quite tain , the end of the platinum wire must be cut off quite clean , and the platinum disk carefully cloaned .
The contact obtained by means of a atinum wirs with a fused end is useless , as its inition is more a mat of prossure than else .
As the apparatus has to be sunk so that its lower end goes to a depth of 5 or 6 feet in water , it is of course necessary to pump in air to distend the india-rubbel ; closing the mouth of the trnmpet , against the .
This implies the nse of a simple pressure , and will also explain the sity for the air-tight glass chamber in front of the fine rubber diahrao.m .
order that the pressure on both sides of this may be normally the same , " " leak\ldquo ; is provided , rullning from the tube supporting the diaphragm to the space in front of it .
This " " leak\ldquo ; consists of a bit of drawn-out thermometer tubing , about an inch lon The apparatus with which all the earlier experiments were per- formed is only suitable for explosions of from 6 or 7 oz. of cotton .
In this apparatus the trumpet is replaced by a drum connected with the upper portion ) the apparatus by means of a tube ving an internal diameter of nearly three-quarters of an inch .
The diameter of the drum was 6 inches , and each drum-head consisted of an -rubber diaphragm , so that each ] gauge had two india-rubber faces .
In order to prevent the india-rubber cutting at the of the drum , the latter must be finely polished round the rim , and this should , if possible , be made of round brass wire .
We found that the decay of the india-rubber was ened by lubricating the of the drum with tallow .
By far the besr to use for the india-rllbber on to the drums is silk '-line .
Both kinds of gauges were fixed to substantial wooden supporfs , and these were provided with copper 0 iron cIcats to run in guides on a thick plank spiked on to the " " pentiue \ldquo ; pile .
The drum gauges were sunk to the required distance by means of lead sinkers ; the trumpet gauges required to be hauled undel water by a block and tackle attached to a 5-cwt .
sinker droppeci at the foot Velocity of Transmission of Disturbances through \amp ; a-uater 503 of the pile .
The handling of these big trumpets required several men ; even the smaller gauges were most troublesome to fix if the weather was bad .
The chief difficulty met with in the earlier forms of gauge was their liability to leak ; in the later forms this was got over by the of tying on the india-rubber and by cemenfing in glass cylindel between two concentric ' of brass with melted .
Give nlixed with bichromate of potabh and melted in the dark is also an excellent cemen , if coated with shellac varnish .
screw the platinum disk ) ssed through a stuffing box filled with an india-rubber bung , which could be jammed by means of a screw top .
During the course of expcriments it was constantly necessftry to adjust the depth of the gauges to allow for the rise and fall of the tide .
the experiments , the results of which are given in this paper , the " " drum\ldquo ; were exc.lusively employed .
This is due to two causes .
the place .
we seldom commanded sufficient manual tance to wurk the tr.umpets properly ; and in the second , these lat er sensitive to the action of small w on the of the water that we could not utilise .
full ivelJess for the pose of registering the explosions .
xcept on few days in the course of the winter , the equisite calmness of the water is not attained in acksoll .
It be mentioned that it is easily possible to seb these to such a point of sensitiveness they will contact on receiving the vel.bal order to do so\mdash ; if it is uttered in the right tone from a distance of 10 or 15 yards .
cormccted the Firing of the The charges themselves were always enclosed in thin tin cylinders , or else in oil-silk or mackintosh .
xcept with respect to the detonators , is little to say .
When were iven to make them for ourselves , we finally , with the assistance of Mr. Cook , hit on the following method .
A wooden mould prepared , of such a form that it allowed a cylinder of plaster of to be cast ound two lengths of -covered copper wire running parallel to each other and to the loug axis of the cylinder .
When the ster had set , wires cut o# at one end to within about lalf a millimetre of end sul.face of cylinder , ) at the othel allowed to protrtlde to a distance of or 6 inches .
fusible of platinum secured to the ends of coppex wire by solder and resin only .
These primed with a good of a mixture of guncotton fluff and meal powder , made into a paste with collodion .
In order to attach the cap , advantaken of the good workmar ) ship of ordinary ca1tridge A mber twelve bore\ldquo ; case and cut off abovo the 504 Prof. R. Threlfall and Mr. J. F. Adair .
On the work in the lathe .
One end of the cardboard cylinder thus obtained closed by having a wooden plug glued into it .
Through this plug a hole was drilled to fit the detonating caps we were able to buy .
A cap was pushed , open end first , into the hole till it was flush with the inner surface of the plug ; it was secured in this position the help of shelIac varnish .
The empty portion of the cap was then iilled with a mixture of fuIminate of mercnry and guncotton flufl , and this was allowed to cover the bottom of the cartridge case above the plug .
Finally , some bees'-wax , resin , and linseed-oil luting was poured over the surface of the upper portion of one of the plaster cylinders , and this was then pushed home , so that the priming rested on tlJe loose mixture above the plug .
Finally , the whole thing was dipped into some of the same luting , so as to give it a roof coat .
The detonators were then tested with a telephone stored .
The advantage of this method is that as each operation is quite simple the detonators can be formed by labour ; a consideration , if many have to be made .
The luting must have some linseed oil with it , otherwise it is apt to crack in cold weather .
We have never known a detonator so ucted to miss fire , though missfires often occurred till we used fulminate and guncotton pJ'iming .
colIodion paste is also very suitable , as it protects the bridge .
In most cases the charges to be fired were contained iu tin cases , big enough to hold one 9-oz .
disk of guncotton .
The lid of each tiIl had a cylindrical tin tube projecting it to a distance of 4 inches ; this tube was not central , because 9-oz .
disks are bored with two holes .
The detonators were secured in these tubes simply by dipping the tubes in hot luting and then pushing the detonators down till the cap projected to a sufficient distance .
The was then placed in the tin , ths detonator passing into one of the holes of the cotton , and then the junction between the lid and the tin was made secure with luting .
For smaller charges .
other receptacles were provided , most cases simply rappinoD the charge in mackintosh cloth was sufficient to keep out the water long enough for our purpose , the joints being well secured with a solution of india-rubber .
One terminal of the detonators was always to earth .
Apparatus for the Mjasurement of the As far as we can learn , it has been usual in the interpret tion of the markings on a smoked plate simply to count the whole number of fork-beats , and to estimate the fractions at best by menns of a pair of spring dividel.q .
In our method is obviou.sly on a par , as .
as goes , with the other mcasurements .
Velocit of Transmission of Disturbances throngh .
505 Besidcs this , the time is estimated by the distance between the breaks , not in one line , but in two lines srated by about 3 mm. , the fork-line being about 8 cm .
below the of the scribing lines .
It was clear at the outset , therefore , that special ements must be made if real accura was to be looked for .
This have attained in mannet :\mdash ; By the construction of the , and especially of the scribers , it is possible to set the two scribing poinls and the point of the fork in ] same vel.tical line .
Now , by proper adjustment , line is brought to coincide with a line passing through the knifeedges and the ceutre of gravity of the smoked surface of the rectangular glass plate when the pendulum is at rest .
The three therefore , at any instaut dul.ing the motion of the pendulum , lie on a radial line passing through the centre of snspeusion of the pcndulum .
If , efore , proper radial lines drawn on the plates , corresponding to ally epoch by the scribcr , these lines will cut ' ' fork\ldquo ; line the ding to the same epoch .
In fact , the object is to find the exact point on the plate which was in contact with the fork at the instant that scriber collcerned to move .
We had a scl.ibing table to find these points in the following manner .
A crossed was prepared , about a foot longel than distance from the knife-edges of the pelldulum to the bottom of the glass plate .
A ground axlc-pin was provided , passing through the board perpendiculal to its plane near one end , and firmly fastclled to ib by brass .
A long brass rod of ular section was furnished with a projection on one side at one end , and is projection as bored the into which the axle-pin was ground .
The brass rod could therefore revolve on the board about a point situated at a distance of about ters of an inch from one of its shortel edges .
The distance between knife-edges of the pendulum and of the glass plate was efully m , and a recess cut in the board such a manner that when the g plate was dropped into the recess it occupied a position with respect to the pivot precisely like upied by it during expe imcJlt with espect to the knife-edges of the petldulum .
A -shap oovo was cut in the upper surface of the rather grenter the breadth of the glass plate .
A efully n plate of brass was furnished with two on its Iowcr fact and with an adjustable ] scribing , just as dicularly from the line joining centre of t'ne studs as the gation of the centre line of the -groove was the centre of the axle .
Consequently , if the brass plate is caused to slide by of its studs up down he groove , tlJe Sibing point marks a Iinc which if produced will pass ) the centre of axle .
The 506 Prof. R. Threlfall and Mr. J. F. .
On the ccuracy of this depends , of course , on accuracy with which the roove is cut , and the accuracy of the equality of the distances between the scriber and the groovs and the prolonged axis of the groove and centre of the axle .
Great care was taken to secure these conditions both in the making of the instrument and in the sting of the scribing point .
The precise estimation in fork-beats of the ce between the breaks in the scriber lines became an easy matter with this table .
The plate was dropped into the recess , and by means of sliding scriber two lines were drawn through the fork-line , each corresponaing to radii of the pendulum and each passing through its appropriate epoch mark on one of the scriber lines .
The estimation of the exact position of ths epoch mark was not easy at first , owing to the slow motion of the scribers .
Finally , however , by making the battery very strong , the springs also strong , the magnets very powerful , and the moment of incrtia of the scribers very small , this difficulty was partially overcome .
Supposing that the two lines ppened to cut the fork-line in exact troughs or crests , nothing more remained than to C0U1lt the fork-beats .
In general , however , this was not the case , and then the procedure had to be adopted .
A sliding-table to a measuring microscope was over the plate , and was to clear the radius bar and scriber .
The scriber was used to draw four lines in addition to the two marked on the plate eady : the so drawn passed through the nearest crests or of the fork-line , between which the line of section lay .
nlicroscope , with a micrometer screw of fifty to the inch the head divided into a hundred parts , was on the table nnd adjusted by a special square , so as to be tangential to the arc of mean position of the fork at the point where that arc is cut by the epoch .
Sometimes the arc was drawn by the radius bar , the correctness of , setting of the measuring microscope was tested by whether the arc left the cross-wires appreciably as the microscope was screwed along .
If all was , three were taken with the micrometer , thus\mdash ; First reading , -wires on line through crest to right Second , , , , epoch , , , , crest to left Q. the whole number of fork-beats has been so counted that an addition has to be made at each end to get the true length , we have : if be the whole time required , the period , the bole number of beats , A and fractions of beats , Let be the micrometer readings as above , then\mdash ; locity of Transmission of Disturbances throuah and if the rate of the pendulum be supposed constant one beat ; in practice this is very nearly attained , for the firing-key is so set as to use only the middle or maximum velocity part of the btroke for scribing .
In some earlier measurements the between successive fork-crests were measuled , with the view of ap the orous correction , but it was fouud that it differed from the one here given by less than the of a fine SCl.iber line on the smoked glass .
It would , moreover , be as likely to be in excess as in defect , according as the fuse was more or less sensitive , and the battery stronger or weaker .
Hence the whole time to be measured comes to\mdash ; In order to eliminate of observatio1l , the plates vere sometimes read by one of us and sometimes by Mr. Pollock , to whom our best tbanks are due .
The measuring microscope by the Cambridge Scientific Instrument Co , , had a respe table screw ; this screw was ith a scale , but fo und to l.equire no correction for its present purpose .
It has been stated th observations were vays made alternately ; first the shot was fired so that it the piles , say , pile X to pile , and then immediately afterwards another similar was fired , so that the shock passed from to X. It is clear on principles that if the time-constants of the and SCl.ibers remain the same from shot to shot , that then , by the metical mean of the times obtained from the plates , ] get the time-value independent of time-constants .
In vcver , to obtain additional security , both the nets , scl.ibel .
S , \amp ; c. , were made as much alike as possible , so as to educe the of their time-constants to the least possible amount .
Our nlethod of elimination , however , is faulty , if is any of time- constant of the es which is reversed by the of the shot .
, for instance , the time-constants of the gauges are of the violence of the shock the ttuges receive , the ) will be outstand- .
errors not eliminated by taking means .
In de .
to test this point , a large number of experiments were made durin , the end of 1886 and inning of 1887 .
here may also be mall differences of velocity , on tion caused by the configuration of the shore 01 sea-bottom .
inally , 508 Prof. B. and Mr. J. F. tdair .
On the and in individual readings , this was often a tiresome matter ; the symmetry of the fork-line was sometimes disturbed by a superimposed small ripple of diiferent period to that of the fork .
This ripple was due to the independcnt vibration of the style attached to the fork .
Of course , as it was small it had no effect on the whole number of beats , but made the estimation of the fractions more difficult and less accurate than would otherwise have been the case .
It is not very difficuIt in practice , however , to estimate from a slightly rippled wave the position of the true crest , and this was always our object in the measurements , as we hoped to be as likely to make them too as too small , and hence to get rid of them in a number of observations .
This unceltainty is , of course , entirely of the second order , and , as a matter of fact , is negligible ccmpared with the irlegularity observed , even if it is not\mdash ; as we entireJy eliminated through the nnmber of the observations .
rperiments to find whether the Time-constant of the Gauges depended on the Energy transmitted .
It is obviously impossible to reproduce the exact conditions of essure in a laboratory , and hence the gauges were tested ith ipect to their time of action by dealing blows of different int , ensity to the india-rubber diaphra , gms .
This was accomplished by means of pendulums of constant lengths but of varying masses .
The arcs through which the pendulums moved lvere also varied within wide limits .
The method of testing was as ollows :\mdash ; The myograph was so anged that the key co1npleted the circuit of the omagnet of a reIay ttus .
The armature of the relay in its position of rest completed the circuit of an electromagnet held the strik pendulum ( by of an iron armature ) a fixed position .
The result of the fall of the pendulum was , therefore , to release the pendulum suspended opposite the gm of auge .
The first SCl.iber was arr.anged in such a manner that it brought into action ectly the bob of the pendulum touched the rubber disphragm .
To this end , the was temporarily with a strip of tinfoil , so that the circuit of the first scriber comprised battery , tinfoil , and testing pendulum .
The second scriber was in circuit with the contact portion of the gauge in the usual manner .
Consequently , on opping the myograph pendtllum , l.ecord of the usual character is obt , on the smoked plate ; the first epoch is that of co1ltact of testing pendulum and tinfoil , he second epoch is that of the completion of the contact .
The pcndulums employed were of different kinds .
One set of experlments were made with a pendulum composed of a light brass rod , on which bobs of diffel.ent sizes and materials could be sc.rewed .
elocity of 'Transmission of through .
509 order to eliminate any effect depending on the velocity of the impact of the pendulum on the rubber , this arrangement was replaced by another , in which the pendulum consisted of bobs supported by fine wire , the armatures being attached to the bobs .
) bobs themselves two in number , one being of lead and the othel of plaster of coated with tinfoil ; the effective length of the pendulumf , iu each case 92 cm .
, the weight of the plaster ement complete was , and of the lead arrar-gement 966 grams .
The arc ough which both pendulums fell was the same , and their periods were approximately equal .
The result of about ten experiments with this atus was that the time-constant with lead bob sec. , and with plaster bob sec. , showing an apparently smaller constant for the heavier blow .
The of the blow was , by the arrangement of the appa ] atus , to masses ; thus the ratio was about as 966 to 160 , or about six to one , and the difference in the time-constants was sec. One wave of the t-fork , however , was rather more thau a centimetre in length , and corresponded sufficiently for the purpose to sec. , and the difference of the time-constants is therefore about sixty-thousandths of this , , less enths of a millimetre on the smoked glass .
This is much too Iarge a quantity to be accounted for by expelimental error in measuring the plates , but is not too large to be for in othel ways .
It has already been mentioned that if surfaces greasy , the definition of the epoch of contact is chiefly a matter of pressure weell the surfaces .
Now the contact with the lead bob of the pendulum was , as far as the lead goes , on .
clean , because the lead was scraped , but with the plaster bob the contact was between } tinfoil surfaces .
Now it was found in the expel'iments above referl.ed to that a much greater pressnre has to be exel.ted to )roduce " " contact\ldquo ; when the surfaces are greasy than when they are clean , and tinfoil is always greasy , and consequently we should expect that , as as this effcct , the time-constant would appeallonger with the bob tl with the plaster one .
This is contrary to what was observed .
On the other hand , tinfoil on the plaster was efised , ) hence , if we suppose that a sensible wave of compression does not to travel np the tube till the deformation of the essed to a certain extent , the effect of the tintoil creases will be to the difference between the time-constants in the same sense as that observed .
We conclude , , that , settling asido the difference in the rate of of the waves in the tube depending on their amplitude , we must look to the part of the apparatus for explanation .
This was ably heavicr than was ultimfttely used in the experiments , and the adjustment was nothing like so delicate as we atter510 Prof. R. Threlfall and J. F. Adair .
On the wards succeeded in making it , and consequently bhere is every reason to believe that the difference in time-constant observed in these experiments was very much greater than it was in the actual observa- tions .
But the times measured are of the order of one-tenth of a second , and the amount we are now concerned with is of the order of second , so it is clear we are discussing the correction of a correction .
It is true , however , thab the instantaneous pressures to which the gauges ars exposed dnring explosions differ by considerably greater amounts than those due to the action of the two pendulums ; it was always arranged , therefore , that the gauges old have a considerable excess of sensitiveness when used to measure the explosion velocities .
It must be added that the length of the chord of the arc through which the pendulums swung in the testing experiments was about 20 cm .
, so the whole time in question is only of the order of the time taken by the pendulums to swing through about half a millimetre { it their .
In addition to this , it was extremely difficult to keep the tinfoil properly and uniformly glued to the sllrface of the rubber .
The effect of blow was always to crumpIe the tinfoil to a small extent ; a wrinkle of the order of mm. would explain the whole effect , and after several impacts much greater deviations were difficult to avoid .
The blows from the lead bob resulting in by far the greatest deformation of the rubber produced much the eatest trouble in wrinkling the tinfoil , and yet we could not rsplace the tinfoil after such experiment ) ecause we desired to be sure that the at ' of the surface as to greasiness was constant , and consequently we were obliged to smooth the tinfoil down to the best of our ability .
It may be added that , of course , the tinfoil was as well cleaned as possible , but it is not a very easy substance to be quits sure about in this respect , and the thinnesb silver and platinum foil we had and tried would not stick to the rubber properly ] stuck by so much marine glue as to weight the diaphragms sensibly , and thus perhaps introduce a peculiarity .
view the sources of above , we did not feel justified in applying any correction to the observed velocities of explosion-effeot on this account ; the effect , therefore , will be that our velocities run the of a too to the extreme extent of , say , 2 per cent. ; we are certain ourselves , however , that this is not the case .
It may be that the experiments with the rod pendulums led , on the whole , to rather smallel .
values for the time-constants , but we hav ' not laid any stress on them , as the pendulnms did not move with exactly equal velocities in all uases .
Velocity of Transmission of Disturbances through .
511 Comparison of Fork with a Clock .
The fork employed was a very excellent and massive one , belonging to a ribbon raph aatus and made by Koenig .
The fork ' was marked 100 .
We felt confidence in this fork from the first , as it was with a style and platinum cord contacb apparatus for eIectric maintenance .
A comparison was made with the laboratory clock ( a very fine one , by Cooke , of York ) by one of us and Mr. Arthur Pollock .
Lord Rayleigh 's method was employed .
The fork compaled directly with the clock was also by Koenig , and was reduced from about 35 complete vibrations per secolld to approximately , by very carefully equalised weights .
The approximate period was found by counting whole and estimating partial vibrations , over a period of 27 seconds ; the raph above alluded to was used for this purpose\mdash ; one scriber rked periods of 2 seconds direct from the clock , and the other was included in the driving circuit of the fork .
The of ribbon used was feet .
We insist on the accuracy of this approximate surement , because the exact comparison requires it .
The clock , with its contact arl'angement , had been carefully rated for some time and adjusted so for this purpose the rate is \mdash ; it was about a second a day .
The observation of the light spot was undertaken by Mr. Pollock ; it was found most conveuicnt to interpose the adjustable slit of a spectroscope between object-glass and the eye-piece of the observing telescope .
The pendulum of the clock made a complete vibration in 2 seconds , and the slits of the opened twice in each complete vibration .
The period of recurrence of the system was 432 seconds , and the fork was losing on The result is that the actual frequency of the is complete vibrations per second .
The second harmonic of this was ) frequency of the driven fork with which the standard was compared .
Hence frequency of fork is\mdash ; The beats were counted over a period of five minutes , the driven fork at C. by a Kew standard and examined ) This tempel'ature was then changed to C. , and the beats counted .
Two sets of experiments were made , which reed very well together .
the observafions at C. the driving-fork eriod of flash recurrence changed to about 270 seconds .
The , the calculation of which presents no new features , is that being frequency at C. , on lbe assumption that the temperature coefficient is constant .
In practice , the lowest temperature on any VOL. XLVI .
.512 Prof. B. Threlfall and Mr. J. F. Adair .
On the was C. This result amounts to making the fork correct at C. An experiment was made to try whether the slight damping of the fork , produced by the friction of the style on smoked glass , caused any appreciable change in the period .
A plate of smok-ed glass was set up so at the style rubbed ths glass , and the beats were again cottnted .
If any effect exists it is less than an alteration of frequency ponding to one complete vibration in two minutes .
The effect on the observed velocity of transmission of explosive energy at the extreme temperatures of and of the temperature coefficient of the fork is and per csnt .
respectively , of tho uncorrected value , and is , of course , included in the reduction of the observations .
Measurement of Distance between the For the purpose of estimating the distance between the piles a spike was driven into the top of each , to serve as a mark for the cross-wires of the theodolite .
No less than three distinct sets of measursments were made for us by Mr. R. J. A. Roberts , Government Surveyor ; was due to the fact thab on one occasion one of the piles was pulled up by a large steamer , which made fast to it in order to swing , and after the pile had been -driven rather badly it got washed to one side by the heavy wash of another steamer .
Finally the pile was moved further away , a better bottom was found , and we had no further trouble .
In the intermediate posihon the pile was not driven quite straight , consequently the distance between the had to be COll.ected according to their depth .
Now the gauges were made to follow the tide , and consequently the COl'rection had to be applied to each ] experiment .
For this purpose , some measurements of the slope were made and a table of corrections for what we called immersion was prepared , by whose aid the distance on each occasion was correctly arrived at .
The gauges were fastened to the piles in such a way that no appreciable correction has to be applied to Roberts 's distance betweell the spikes .
We do not mean to say that there may not be some small difference arising from the thickness of the drum ( about 3 cm .
) and accidental straining of the woodwork , but it cannot amount to more than a few centimetres , and will vary from day to day , a.s probably in one direction as in .
The piles were also very possibly a little bent by the wind and tide ; the amount of this could nob be estimated without a theodolite permanently set up , and this we could not do , partly because we had eodolite of our own .
However , this small error is as likely to be in excess as in defect .
Much more important for the absolute velocity is the ] acement of the firing buoy , by wind and the direct line of bhe piles .
Velocity of Transmission of Disturbances through Sea-wafer .
513 buoys wsre anchored with sinkers and attached to the pile by opes , so as to keep them as rigidly as possible in one position .
However , at low tides the ropes became slack , and then if a wind was blowing , the shot drifted for a few yards out of the direct line ; the result will be that the second gauge may be affected by a portion of the wave front ( if there is one ) which is not identical with the portion affecting the first .
Owing , however , to the end of the mooring line being permanently fastened to the nearest pile , the shot displaced will move in an arc of a circle , with the pile as centre , and hence we have a meaus of the maximum effect of any amount of drifting .
It can readily be seen by drawing the that drifting of the shot will have the same fect as diminishing the distance between the piles , and hence the times will be rather shorter , and the velocities rather greater , than would otherwise be the case .
An error of this kind will not , therefore , be eliminated by any nuolber of observations , as it is always in one direction .
The emedy is to take as much pains as possible to prevent its occurrence , and this was always done ; an account of the magnitude of the which may arise in this way is contained in the third part of this paper .
METHOD ADOPTED To THE DISTANCE BETWEEN 1N PILES DRJVEN 0UT , BERRY'S BAY , PORT JACKSON .
The base-line was measured with a link -inch steel riband with adjustable ends , to which was applied a tension equal to 16 lb. The chain was found after careful testing to be equal to the standard at the Government Observatory , Sydney , at a ) ature of F. It was tested on a calm , cloudy morning , the temperature being even , and the base-line was measured the same day , the temperature .
at the time F. The standard at the Observatory is of an inch short , it having been measured off with the compensating rods for measuring trig.-base lines .
A correction was therefore applied to make up for this error .
The chain when tested was supported bhrou .
out its entire length , but was suspended when used measuring the base-line .
By careful experiment it was found that this made a difference in length of of an inch , for which a correction was also applied .
The expansion or contraction equal to of 1 inch for F. for one chain , the distance in chains , multiplied by half the difference een standard and observed temperature , and the result dividod by 8 , gives a correction in decimals of a link to be added or subtracted , as the ature is or lower than the standard temperature .
To the handle at the back end of the chain was attached a small 514 Prof. R. Threlfall and Mr. J. F. Adair .
On the piece of steel with a , and this was regarded as the end of the chain .
This end was worked by means of a slow motion screw ched to a board , on which the chainman knelt , and so the end of the chain was kept perfectly steady .
The was brought by the screw the mark , which was a very fine pencil dot on a piece of white paper , gummed on to a nail driven into the ground .
On the forward end of the chain was hooked a spring balance , which indicated a .
strain , applied by means of a light straining pole , 12 feet , thus the chainman to apply the strain far more steadily and easily than by any other method at present in use ; and also allowing the full length of the chain to be used on a slope equal to about 1 in 5 , the end of the chain up and down the pole .
The greatest slope in this case was , however , only about 1 in 10 , aild only for two chains of the entire length .
The forward end of the chaiin was as the position of a small hole , through which a string , with a small plummet attached , passed .
This plummet was used merely to indicate the position of the ilail , and to keep in adjustment a contrivance which gave the position of the nsit theodolite ( carefully adjusted for the telescopic axis ) at right angles to the ction of the chain .
The end of chain was sighted by means of the transit instrument , and the telescope being then depressed , the fine point of a hard pencil was lined on the paper before described .
This method does away with any errors due to the use of the plummet ( especially in windy ther ) for the end of the chain on sloping ground .
By the use of the instrument the end of the chain can be marked to of an inch .
This method has also other advantages accurate measurement .
The instrument used was a 5-inch transit theodolite , by Troughton an 1 Simms , of London , having the circle graduated to 20 seconds .
The teleseope was larger than , and of much superior , those usually fitted to 5-inch instruments .
The mmet was suspended from the end of the vertical axis .
The instrument was used on a traversing top stand , and being in very good order , and the axis moving freely , there was no twisting pressure ( torsion ) on the stand , the day being , cloudy one , and the sun completely obscured , the cffeet of sunlight on the eyes was avoided .
The instrument was careilllly set up over each of the points , and on the base-line , were observed to the spikes in the piles ( points Fi and F ) off line A. The angles shown on the plan are the means of four at each of the stations .
After the first two readings the tclescope was turned over in the reversed position , so that ally .
possible errors due to change of focus ( movement of the optioal axis of the instrlllnent ) , of horizontality of the telescopic axis , and impercollimation , are entirely eliminated .
The levels were sensitive , and of good quality .
readings of both verniers were recorded Velo of Disturbances through Sea-water .
515 and a mean taken , and each was lead off a different part of the graduated each time , so any duc to icity was entirely iunted , and any errors due to imperfect yraduation vere reduced to a minimum .
In an on the graduated circle , not only is the second measures of the given on a diffcl.ent part of the circle , but all angular quantity , which pel.haps the vel.nier is incapable of showing , has been bled , and ma therofore become appal.enl .
As an example , suppose the itt efully sct over the point ( see Plate 3 ) and levelled , and itial 1oint , the two niers a read off , and their ition ecorded i tho ]dbook .
Leaving the lower clamp fast , and oleasing the screw clalnping the verllier plate and that the , the point Fi is then obsel.ved to , and the ecorded , so on to the F. Then ' upper } ) { thd the lower , the point is referred to , and the tion repeated three times , thus meftsllres the a after the second roulld to everse the telescope .
This movement will the vernier originally on the left hand side of the to the right hand .
A mean of the verniers and all the eadings is the .
Tripods with steel points and hcavy plunlmets were used as ence marks fully set the ends of base-line .
The alignment of the points and was carefully pel'formed , using the telescope in reversed positions , both the ends of and at the ) themselves .
The triangles ined by the above operations for hxing and were too small to make it necessary to go inbo the equation of conditions .
\mdash ; REDUCTIONS OF AND ALCU L Oh THb OF SOUND rfHROUGH Distance over which Distnrbcmce Travelled .
The distance between the gauges was at first 15,410 cm .
, it afterwal.ds 15,840 cm .
, minus a correction made for slope of pile which had been driven slightly out of the vertical .
This tion varied een 51 cm .
and 59 cm .
, and depended 011 the depth of the below the top of the pile .
inally in the third position the distance was 18,210 cm .
Time Jleasurements .
Let be the epoch of alrival of the disturbance at and Iet be the time constant of the and the scriber attached to it respectively , then the record 011 the smoked plate occurs at 516 Prof. R. Threlfall and Mr. J. F. Adair .
On the time .
Similarly if be the time required for tfJe transmission of the disturbance , then the second record on the smoked plate occurs at time , where are the time-constants of the second gauge and its attached scriber respectively .
Thus the observed interval as recorded on the smoked plate is\mdash ; Similarly for a disturbance the opposite direction the recorded interval is\mdash ; provided that the time of transmission is the same in both directions , and that the time constants of the gauges and scribers have not altcred between the two observations , and that the scribers are attached to the same gouges in the two observations .
Hence time of transmission of disturbance .
The velocity of transmission calculated on the supposition that the has an exact frequency of 100 must be multiplied by , to get the correct velocity .
The clocities had been original ] deduced on the supposition that fork was correct ; these uncorrected velocities are given in Column 9 of the Table ; the correction arising from the decimal parts of the above factor , , from , are Column 10 , and the corrected velocities in Column 11 .
The complete formula for the velocity is thus\mdash ; where is he distance between the gauges and is the temperature of the fork .
This gives the velocity at the temperature of the seawater duling the servation .
Theoretical Calculation of the elocity .
In Column 12 of the Table is given the velocity of sound , as follows .
The formula giving the velocity of sound through water is to be\mdash ; where is the adiabatic resilience of volume , and is the density at { he temperature considered .
Velocity of smission of Disturbances thronqh Sea-water .
517 Observ and Calculation of the Density .
the " " Challenger Report vol. 1 , 'Physics and Chemistry , ' p. seq. , is given a table calculating the of a seawater ab any temperature when its density at C. is known .
get the density of the 's Bay water at this temperature six eriments were made on two samples collected at an of a llonth .
water of the Bay is practically Pac.ific Ocean water , being to all intents and purposes no fresh water into Bay .
Four of the experiments made by .
a piece ot glass flint , both in doubly vacuum-distilled water and in the ater ; and two experiments were made with a spccific rravity bottle .
Both the lass and the water had been to stand in the balaIlce case .
seven hours before weighing ; the thertlometc.r uhed was a standard at Kew , the ction beilJg , it happened , , for the temperature observed .
The picce of glass was from the arm of balance by a platinum wire , the Iengfh of was 28 cm .
of 1 metre of the wire was , and the density was taken as .
In each hing in wnter one half the wire was .
The sensitiveness of the balance was such ve leadings of scale di isions per .
The of the sea-water is compal.ed with of by means of the which is the density of the water in which the glass is mass of lass , of lass , apparent weight in rent weight in water , .
of the air cighing , density of the mass of rram , density of platinum on sido is above got by ilh Tn esent case this is equivalent to neglecting\mdash ; in rison grams .
Prof. R. Threlfall and Mr. J. F. Adair .
On the By applying the above formula to the weighing in sea-water at C. and to the weighing in pure water at C. , the first expe- liment gave density of sea-water ab density of pure at .
The density of pure water at the various tempel.a- tures was taken from Kupffer 's obselvations as reduced by Professor W. H. Miller ( Everett 's ' Units and Physical Constants , ' 2nd edition , p. 39 ) .
The densities for ures intermediate to those given by Everett were obtained by ion .
For the measurement of the density by means of the specific gravity bottJe , two bottles were used , one acting as a counterpoise , together with the small requisite quantity of sand and minium foil .
The other bottle being quite clean was then filled with the water to be , great care being taken to avoid the retenlion of any bubble of air .
The dry and wet bulb thermometer was observed , together with the barometer , in order to make the usual colrections for weighing in .
The results of the two measurements made by this method agreed closely with the other four measurements made by weighing glass .
In to have comparable esults and to get a mean , the results of the measurements made at , and C. respectively were reduced by " " longer Reporb ' Physics and Chemistry , ' vol. 1 , p. 70 , to the temperature C. These comparable results are given in the following table .
Weights were assigned to the observations shown .
Hence the probable alue of the density at C. is grams per cubic centimetre .
The densities of the water at the various temperatures of the experiments the Bay were thence found by means of the table in the " " Challenger Report The temperatures of the water in the Bay during the experiments were\mdash ; weights assigned to observation are because the weighings taken at midnight .
520 Prof. R. Threlfall and Mr. J. F. Adair .
On the Here is the absolute temperature and is Joule 's equivaJent .
The equation be written\mdash ; in which is the coefficient of expansion , and with a negative sign is the isothermal compressibility , or the reciprocal of the isothermal resilience .
Thus , if coefficient of expansion , and I3 the density which is put for Thus it is requisite to , ffie specific heat , besides the quanbities represented by the other letters .
With a view to obtaining , it was discovered from Watts 's Dictionary of Chemistry , ' vol. 5 , p. 1017 , that the composition of Pacific Ocean .
at a depth of 11 feet , Lat. S. , Long. W. , is\mdash ; lnaking the total solid content per cent. approximately .
Again , in vol. 7 , 2nd Supplement , p. , it is stated that a solutio has a specific heat , with a specific ; and this solution contains approximately per cent. solid matter .
Thus , since occurs in a small term , it may be taken much.error .
An experiment made by Mr. Flint , one of the students in the Laboratory ; gave a value practically the same as the value here taken .
The value of has been previously found to be is taken , and the density given in the Table .
Velocity of Tr.ansmission of Disturbances through Sea-water .
521 of the Resiliences .
With regard to , Professor Tait has kindly supplied us with information .
He gives formulae for the compressibility of fresh wnter and of sea-water at low pressures and high pressures .
The compressibihty , as given by him , is a function of the pressure and the temperature , and thus for a pressure which rapidly from a low to such a value as one or three tons weight per square inch , the compressibility would be a variable quantity .
It did not ppea at first his formula for moderately low ures , or the mula for such pressures as from one to three tons per square , was to be used to get the reciprocal of the sibility .
The further uncertainty as to the effect of viscosity is not allowed for in the theoretical velocity .
With a view to settle of Tnit 's formulae was to be adopted , it was remarked that for an liquid , subject to impulsive pressures , the equation determine the impulsive pressure at any point is And ence , for an impulsive pressnrc uniformly distributed over a , the imp lsive pressure at a } ) oint outside the sphere would be to the distance at that point from the centrc of It seems desirable to determine the function of , and , which satisfy boundary conditions and the equation from the of continuity combined with the tions of hen a relation is assumed between and .
If it is posed that impulsive pressurc is subject to the of ordinary a relation as thnt given by Van der Waals pel.haps be taken , or one of Tait 's relations giving the compressibility as a function of iure and temperature .
upposing that the abov ) considerntion applics to atel , it was further concluded , by of elot in cussing t experiments of rrau and Vieillc , in our the initial on ) walls of the case containing the explosive was in each case ( guncotfon and dynamite ) about 8000 , rams .
sq .
cm .
This ement between the initial pressure due to guncotton and due to dynamite is accounted by the density of boing rather differenf in ] two cases .
Supposing this sure uniformly distributed over a sphcre of 1 foot diameter , Prof. R. Threlfall and Mr. J. .
Adair .
On the which is very much greater than the volume actually occupied by the explosive , and that the pressure at a distant point is inversely proporbional to the distance from the cenbre of the sphere , the pressure at 20 yards distant would be approximately atmospheres of 1,014,412 dynes per .
cm .
each .
This value for an atmosphere has been taken as representing Tait 's value , which he defines by atmospheres ton weight per square inch , at burgh ( .
For below this , Tait 's low pressure formula is applicable .
If the sphere supposed to be occupied by the explosive is less than 1 foot in diameter , the pressure at a distant point would , on the tbove suppositions , be proportionately diminished ; and at greater distance than 20 yards it would be further proportionately diminished .
Viscosity would aid in further diminishing the pl'essure , so that on the whole , for the space between the gauges , it seemed advisable to use the low sure formula as given by Tait .
This ] is\mdash ; Average compressibility of sea-water at low pressures is per atmosphere increase of pressure at temperature atmo- spheres being ton weight per square in Taking such an atmosphere as 1,014,412 dynes sq .
cm .
, we thus get\mdash ; The following table gives the values of this for the temperatnres of the observations .
General Formula for the Velocity of Sound .
The on p. 516 thus becomes\mdash ; Velocity .
; Velocity of Transmission of through Sea-uater .
523 or , substituting the values already found for sea-water , Velocity As stated before , the velocities thus calculated are given in column 12 of the general table .
The calculation is laborious .
Explcmation of the General Table .
The explains itseIf exoept with regard to the observations on July 11 , and the three first on August 10 .
In these cases a of three intervals was taken by taking a mean of two shobs the direction left to right , and then a mean between that mean and the interval from right to left .
It is to be remarked that in the case of the pair of shots on September 13 , which gave the ean interval , the mark on the smoked platae determining ] interval was extremely faint .
A photograph of this plate is show ] ] ( p. 524 ) , the single radial line showing the slight break in the line traced by the scriber .
On the lower part of the Genera.1 Table are given the observations which a mean interval could not be , together with the first observations on April 25 and May 7 , and another pair of observations on July 5 .
These observations were rejected on account of the gauge having water-logged ; besides this they should be regarded with suspicion , since thsy would give resuIts hose departure from the final mean would be greater than the " " maxlmum error\ldquo ; found in accordance with the of adjustment of observations .
It may be added that as a general resuIt of our experience we found that ( possibly ofving to some interference effects ) the distant gauge was often more violently affected after its first indication .
Now when the ganges were water-Iogged the .
becamc deficient in sensitiveness , and consequently in the observations referred to the probability is that the mark observed corresponded to the second , not the.first shock .
This conclusion is cngthened by a reference to the photograph of the plate of September 13 .
PARr III.\mdash ; REDUCTION AND DISCUSSION sULTS .
The experiments are divided into four classes , according to tbc nature and quantity of the explosive used .
In Class A the explove was a .
disk of guncotton , in Class oz. of dynamite , in Class oz. of guncotton , and in Class .
of guncotton .
In Prof. R. Threlfall and Mr. J. F. Adair .
On the CIass A the velocity seems to increase with the temperature , but no such law was de ( iected in the other classes .
In order to reduce the observed velocities ( when corrected for the temperature of the fork ) to a comparable state , they are reduced to one temperature ; and in the absence of a rule for doing this it is supposed that each velocity in Class , \amp ; c. , is expressible in the form , where denotes the velocity at some between the extreme temperatures of the observations in CPass and is the excess of the temperature of the observation above the chosen temperature .
For A the emperature chosen was the meau of the temperatures of the observations ; this mean is approximately C. Hence arises the system of equations If the method of least square , be considered applicable , the equations for the probable values of and are ( Stewart and Gee , vol. 1 , p. 274 ) : .
These give\mdash ; This large value of will be merely used in getting the probable error of the mean , , of the above value of .
If it has a physical meaning it is noteworthy .
Substituting the value 25890 for in the foregoing equations there arise the values : Mean as above .
By Bessel 's formula , probable error of mean By Peters 's formula , probable error of mean tho result for Class A may be written\mdash ; velocity at C. metres per secon The theoretical velocity of sound is 1523 metres per second .
Treating Class similarly , the mean temperatul'e of the observdtions is C. convenience of calculation is taken to the velocity at , and ths system of equations , :\mdash ; elocity of Transmission of Disturbances through .
529 Thus the result of Class may be expressed\mdash ; Probable value of velocity at metres per second .
The theoretical velocity of sound is 1508 metres per second .
For Class the mean temperature is C. Using to represent the velocity at , and being a quantity similar to the quantity in Class , the system of resulting equations is\mdash ; Hence probable value of 194192 , Substiluting this value of above , the values are\mdash ; If the method of least squares be applicable to such a small number of values , then probable error of mean ( Bessel ) ; or by 's formula probable error of mean .
Thus the result of Class may be written\mdash ; Probable velocity at C. metres per second .
The theoretical velocity of sound is 1.525 metres per second .
For Class , reducing the values to C. , the equations become\mdash ; Hence probable value of The values of are therefore\mdash ; 530 Prof. R. Threlfall and Mr. J. F. Adair .
On the Mean as above Departure from mean .
-8716 Hence probable error of mean ( Pefers ) ( Bessel ) \ldquo ; Thus the result of Class may be written\mdash ; Probable velocity at C. metres per second .
The theoretical velocity of sound is 1528 metres per second .
Thus in each class the experimental velocity is greater than the theoretical velocity of sound .
During the experiments it sometimes occurred that the firing buoy drifted slightly ( by influence of wind and tide ) from the line conning the piles .
The deviation from that line could only be approximately estimated , but it was judged that it was never so much as metres , although it was usually slightly out of line .
ppose its distance from the line of the piles to be metres , that the deviation at the other end was the same , then taking the distances as marked , A and being the the firing point , if the explosion occurs at time , and if be the velocity of transmission of the disturbance , this reaches A at time and reaches at time , and the first interval recorded on the smoked plate is The second interval of the shot from ' is Hcnce , if AP tocity of Transmission bances through Sea-water ) AP and V .
But the velocity , calculated on supposition of the ' point being in line , is BP\mdash ; AP From the and the true velocity V is less the velocity calculated in the General Table by per cent. of velocity in the Table .
his amounts to\mdash ; For the mean of Class , viz , metres , the correction is metres .
Suppose the firing point distant metre out of line , a similar calculation shows the correction to the velocity in the Table to be of that velocity or per cent. For the mean velocity of Class A this correction is It is probabJe that in general the firing buoy was not anything like a metre out of line , and hence it is clear that it is useless to apply ] correction to the observed velocities .
In any case a glance At the able will show that the irregularities observed are of such an rder render any attemI ) to adjust the observations in this of 1 ) urcly itio us vii.ltlP .
532 Prof. R. Threlfall and Mr. J. F. Adair .
On the DISCUSSION or RESULTS .
It will be convenient to collect here the main results of the investigation as far as the comparison with the velocity of sound is concerned .
Though the regularity of the mean resuIts is very satisfactory , glance at any of the Tables will show that several individual obser7ations deviate to the extent of nearly per cent. Now from our apparatus , and from the fact that the observations are to a great extent made automatically , we are at a loss to account for these deviations , unless they are real .
We hope to be able to show the cause of this immediakly , for the moment we wish to add a little to what has already been said as to the rejection of certain observations .
Whcn we first began to get readings onr gauges were not nearly so satisfactory as they afterwards became ; the sensitiveness was sometimes so small that we occasionally failed in obtaining any record whatever from the apparatus which was furthest away from the firing point .
The gauges were also apt to allow water to leak in at the bottom , and the air sometimes escaped slightly at the top , allowing the rubber faces to collapse .
Both these accidents tend to lessen the sensitiveness of the gauges very materially .
It was soon noticed that the abnormal resuIts\mdash ; and these always occurred in the direction of making the velocity too small\mdash ; were obtained hen the gauges were for some reason or other more or out of order .
When , as shortly occurred , we succeeded in making the gauges uniformly sensitive , and preventing water-logging and the escape of air , we got no abnormal times .
The only difficulty we experienced in deciding to reject certain obser* Each experiment requires two separate explosions and time-measurements .
Velocity of Transmission of Disturbances through Sea-water .
5.33 vations lay in the faot that at first we could see no reason why , if the gauges worked at all , they should give results which differed .
We could not see that the introduction of water , for instance , between the rubber faces ought to make the time-constant of the gauge greater\mdash ; and yet at first we could only explain the abnormal low velocities by supposing that the time-constants were enormously increased at the further stations .
A fortunate observation , however , on September 13th , put us on the right track .
We had noticed on several occasions that the first mark made by the scriber on the smoked plates was not nearly so strong as the second or third .
These marks did not depend in any way on the natural period of vibration of the scriber , for that was far too short , and hence we put them down to the passage up and down the air-tube of the gauge of pressure waves .
It was not yet clear , however , why it was that sometimes the second and third marks of the scriber were stronger than the first .
The strength or distinctness of these marks depends on the conductivity and duration of the contact effected in the gauge head ; and this must depend on the increase of pressure and on its duration in the water in the neighbourhood of the rubber disks .
Now it is probable that several waves and not a single wave of compression start from an explosive centre .
First there is the sudden expansion by the explosion , then the cooling or escape of the gases and the consequent falling together of the water before the disturbance subsides .
A glanoe at the map will show that our gauges were so with respect to the sheer stone wall of the quay that interference might well be looked for .
then that interference takes place at the there gauge between the direct and reflected waves ; it is clear that the first pressure to the gauge is exposed is not necessarily the strongest , and equally clear that the duration of interference pressure may be longer , but cannot very well be shol.ter than that of the pressure first arrivinr .
It may well happen , therefore , that the second or tbird marking on the plate is the most distinct , or , even if the sensltiveness of the is low , he first mark might be suppressed entirely , and the secQnd appea ] ' in its stead .
Now , the distance from the first to the second mark on the plate was pretty constant so as the conditions of } ) experiment remained the same , and , strangeIy enough , as it ) seemed to us in the case of the abnormal results referred to , there was also a certain reement .
It seemed as if the velocity either a good deal greater than the velocity of sound\mdash ; or else a deal less ; the abnormal velocities were consistent with themsel in a rough way .
On September , however , we found out the \mdash ; on that day we got a record in which the first motion the scriber ( giving a normal time ) all but too small for detection\mdash ; Prof. R. Threlfall and .
J. F. Adair .
On the a little further on there was a strong mark giving one of tho usual abnormal times .
The esult was that we obtained a normal observation , but we saw that had the gauge been slightly less sensitive , or the explosion htly smaller , we should have had an abnormal observation .
In order to obtain every satisfaction on this point , we applied ( the reduction of the observations ) the mathematical criteria of rejection , found that we should thus reject those observations \mdash ; and those only which we had eady decid to reject on experimental lnds .
Altogether we rejected three experiments , involving six sbots , out of forty-six involving ninety-two shots .
We now pass on to discuss the smaller variations in the observations ich we retaiu .
Jt will be remembered that at first onr primary in undertaking this work was to find whether there were any ities or not , with a view to finding whether the directed action\ldquo ; already commented on existed to any great extent .
If it did , we hoped to find considerable deviations in the velocity , incipally above the mean .
The result has show1l , however , that while such irregularities do tainly exist to a slight extent , we have no evidence to show that directed action takes place from the explosion of freely suspended idges .
The resuIt would probably have been differeut if we had endeavoured to produce such directed action by the application of perly disposed obstacIes or air spaces round detonating pointWe think that the small deviations , as they are , are real , and do not depend on the gauges or chronograph , and certainly not on our measurement of the plates , which was always pushed to a much of certainty .
It seems probable to us that though the cted action is small , it exists to some extent .
The observations emselves afford proof of the fact that waves of great energy faster than sound , and , indeed , in free water , at all events ( and , of course , we have no opinion beyond ) , the velocity is greater the eater the amount of energy transmitted .
If it happens , therefole , that iu any experiment the direction of reatest actio1l coincides with the line passing the shot through the two stations , we shal ] that explosion register an mally h velocity .
On ) contrary , if the energy escapes away from the piles , we shall get a eIocity co1Tespondingly low .
In intermediate ections of escape , th ' velocity will be termediate .
A somewhat delicate question presents itself as to the precise method by which these big waves may be considered to become extinguished .
A preliminnry question which might very well be asked would be as to ether the inc ] in the is to be for as depending on a greatly increased adiabatic esilience .
do not think that any probable in the resilie1lce would of Transmission of Distu.bances th rough Sea-water .
account for otlr resuIts\mdash ; nor , indeed , would this be consistent with the remarks made on the subject of choosing Tait 's formulae .
On the other hand , we have reasons for thinking that after the ressure increases beyond a certain point , the resilience may increase with very great rapidity ; if this is the case , it will explain our resuIts , the wave would rush past the first gauge , and then slow down with comparative suddenness .
It may be remembel.ed that with respect to Class , there was some evidence in favour of thinking that the velocit .
depends largely on the tempel.ature of the water ; this conclusion was not borne Class the temperature was slightly higher .
Without to say that the evidence advanced is of any real importfince , depending as it does on single observations encnmbered with their private peculiarities , we may note that it would not be at all unlikely for veJocities msured as we measul.ed these to have ] arge positive temperature coefficients .
There is every reason to suppose that under the conditions of our experiments the viscosity of the water will be an important factor in rate of decay of the disturbance as it is propagated outwards .
Now , of all the physical properties of water , viscosity is the one which varies most lapidly with tempet.ature , and , consequently , it is not unlikely that he decay of amplitude , and hence velocity in the disturbance , may depend to a gleat extent on the temperature .
In addition to the wave of great amplitude whose velocity has formed the subjecb of this paper , there are , in all probability , waves of varying degrees of amplitude and velocity from the explosion .
These waves , together lvith the final groul ) , the velocity of sound do not , at present , present any features of parlicular terest

