]\gt ; A Determinaiion of the Ratio of Mass to Weight for a Radioactive Substance .
By L SOUTHERNS , .
( Cantab .
) , B.Sc. ( Lond. ) , Whitworth Scholar , Exhibition Scholar , Emmanuel ( Communicated by Sir J. J. Thomson , F.R.S. Received June 30 , \mdash ; Read , 1910 .
) Experiments have been made from time to time by different ervers .
with a view of testing the con stancy of the ratio of mass to weight for various substances .
Their results have been uniformly negative so far as any deviation from constancy is concerned .
Several years an experiment of this nature was undertaken by Prof. Sir J. J. Thomson , who used a pendulum the bob of which was made of .
The quantity of radium , however , was small , and it was found to be impossible to obtain a very high degree of accuracy .
The considerations which led to the supposition that radium might differ from non-radioactive substances in the above respect were as follows:\mdash ; * " " The simplest electrical system we know , an electrified sphere , has attached to it a mass of ether proportional to its potential , and such if the mass were to move with the velocity of light its kinetic equal the electrostatic potential of the particle .
This result be tended to any electrified system , and it can be shown that such a system binds a mass of the ether proportional to its potential energy .
Thus a rt of the mass of any system is proportional to the potential energy of the system .
" " The question now arises , Does this part of the add anything to the weight of the body ?
If the ether were not subject to gravitational attraction it certainly would not , and even if the ether were ponderable we might expect th as the mass is swimming in a sea of ether it would not increase the weight of the body to which it is attached .
if it does not , then a body with a large amount of potential energy may have an amount of its mass ) a form which does not increase its weight , and thus the weight of a iven mass of it may be less than that of an equad mass of some substance with a smaller amount of potential energy .
Thus the of equal masses of these substances vould be different .
" " The radioactive substances are constantly giving out large quantities of heat , presumably at the expense of their potential energy ; thus , when these substances reach their final non-radioactive state their potential energy must * Prof. Sir J. J. Thomson , Presidential Address , British Association , Winnipeg , 1909 .
weight for radium would be greater by about 1 part in 13,000 than for its 4 non-radioactive products These considerations also apply to uranium , for " " we have got good reason for that uranium is a parent of radium , so that the great potential energy and large ethereal mass possessed by the radium will be also in the uranium In the present paper some experiments made with uranium oxide are described .
The results , however , show that the ratio of mass to weight of this substance does not differ appreciably from that of the non-radioactive substance with which it was compared .
The paper is divided into the following sections:\mdash ; 1 .
Preliminary experiments with a wire pendulum .
2 .
Theory of a new rigid pendulum .
3 .
Description of the apparatus .
4 .
Method of experiment .
5 .
Experimental results .
6 .
Conclusion .
1 .
with ( a Wire A pendulum based on the principle employed by Bessel in his investigations on the value of for various substances was constructed for Prof. Thomson by the Cambridge Scientific Instrument Company , and observations were made upon it by the writer at Prof. Thomson 's request .
Although sufficiently accurate results were not obtained with this apparatus , it may be briefly described , as it was of great service in shaping the subsequent course of the investigation .
It consisted of a hollow cylindrical bob of aluminium , which could be suspended by a wire from a by means of a suitable attachment .
Either of two wires , of lengths 75 and 150 cm .
respectively , could be used .
The method of experiment was as follows : bob was filled with red lead and suspended by means of the long wire and the 1910 .
] to Veight for a Radioactive Substance .
time of swing deterluined .
Then the short wire was substituted for the long one and the time again observed .
The red lead was then replaced by uranium oxide , and the times taken as before .
The pendulums approximate to simplependulums .
Assuming them to be actually simple , and writing , for the times of the long and short pendulums respectively and , l2 , for the lengths when the first substance , for which the acceleration of ravity is , is employed , and the same letters dashed for the corresponding quantities for the second substance , we have .
But difference in the lengths of the wires irrespective of the absolute ] , and therefore so that the ratio of the values of is deduced from the times alone .
In the actual case , since the pendulums are not simple , it is necessary to know approximately the distance of the centre of ravity from the in each case , as well as the other dimensions of the appal.atus , errors in the packing of the bob , with corresponding displacements of the centre of gravity of the substance , will have but little influence on the accuracy of the nents .
After many experiments had been made with this ratus , and considerable experience ained on minor points , it was discarded , mainly because it was found to be impossible to fit ether the different parts of the pendulum always in exactly the same way , so as to give sufficiently consistent results .
Thus differences of 1 in 10,000 or 20,000 frequently occurred in the time of swing after all other errors which could be detected had been as far as possible eliminated .
In view of this , it appeared desirable to the writer to construct a rigid pendulum with two knife-edges permanently fixed to it , and to employ some kind of lifting apparatus by means of which the from one knife-edge to the other could be made without in any way handling the pendulum , or opening the case in which it was enciosed .
Such an apparatus .
was constructed in the laboratory workshop to the writer 's designs , and a standard invar pendulum was supplied by the Scientific Instruntent Company for the purpose of checking the clock which was used in the experiments , as the rate of this had been foumd to be somewhat variable .
The theory given above is of course inapplicable to the new pendulum , but a method of the results in a simple manner will be given in the next section .
The only vback in the use of a rigid pendulum is that The pendulum consists of a rod AA ( fig. 1 ) , having two knife-edges fixed at the places shown in the figure , and a hollow cylindrical bob which can be filled with either of the substances to be experimented on .
The bob is not permanently fixed , but can be moved through a small range up and down the rod .
By means of suitable arrangements the time of be taken first about one of the knife-edges and then about the other , the position of the bob remaining unaltered .
Considering the time of swing about one of the knife-edges , we have , where I is the moment of inertia about the knife-edge , the total weight of the pendulum , and the distance of the centre of gravity from the .
We shall assume that gravity does not change during the course of the experiments , but may be different for the different substances used .
total weight , however , of the pendulum remains constant through all the experiments , for the weight of the bob is always carefully adjusted by means of a balance after being filled .
The balance of course ensures ] of weights and not of masses , the ordinary use of the balance for the determination of masses depending on the supposed invariability of for the substances employed .
Since therefore in the case of uranium we know neither mass nor acceleration of gravity , but only weight , it will be convenient to write the expression for the time where is a constant and is the total weight of the pendulum .
iations of mass will only affect the moment of inertia in this expression .
It is possible to fill the bob so uniformly that the centre of gravity shall always be exactly at the same point , but it will be seen that errors due to this can be rendered unimportant by the method to be explained .
We will first , however , assume that the bob is uniformly packed , and that a normal substance is used so that has its ordinary value throughout the pendulum .
It is possible to find a point X , say , on the rod about which the time of swing of the pendulum is independent of small changes in the position of the bob .
In practice , the knife-edges are fixed one above the point X and one below it .
The time of swing of the pendulum about a point of the rod above X will be increased by a slight lowering of the bob .
The time of swing about a point below X will be diminished by the lowering of the bob .
Let , be the times about the upper and lower points respectively , the bob bein in its fl ff 1910 .
] Mass to Weight for Radioactive Substance .
329 first position , and , the times with the bob lowered .
Then and .
This is shown on the accompanying small diagram , in which the ordinates represent times from the upper point and abscissae times from the lower point .
: .
DIAGRAM A. RAlf B. Points corresponding to other positions of the bob will lie on the line 1 , which will be nearly straight if the displacements are very small .
This line will be called the characteristic li of the pendulum for the given points of suspension .
We will now return to the expression for the time of , and suppose it to apply to one of the observations on the diagram .
Suppose the mass of the bob to be increased by the use of another substance , the weight remaining unaltered .
Now I will be increased , and therefore will be greater than before .
This is true whatever point of suspension be used .
If the bob be replaced in exactly the same position as it was when point 1 was taken , and the new packing be exactly uniform , the new point , 3 say , will be higher up the diagram than 1 , and also more to the , both its co-ordinates having been increased in magnitude see small .
If the bob be now moved down to the position which it occupied when point 2 was determined , a new point 4 will be obtained on the line approximately parallel to 1 , 2 .
Thus the pendulum will have a new characteristic line nearly parallel to the old one .
In order , , to find out whether equal weights of different substances possess different masses , it is only necessary to plot a few points on the diagram for different positions of the bob for each substance , and to draw the characteristic lines .
If the masses are different , the lines will be separated ; if not , they will coincide .
Accidental variations of the height of the centre of gravity of the bob , due to inequalities of packing , merely cause displacements of the points the characteristic lines , and are therefore of no consequence .
It is necessary , however , to exercise care in packing , as otherwise later , and it will appear that they also are unimportant .
Corrections for changes of temperature of the pendulum and of the density of the air will be required .
As the experiment is not intended to give absolute , but only relative , results , it will not be necessary to reduce to zero of temperature and density , and therefore we shall only consider deviations from the mean conditions .
Thus the corrections will be small , and an approximate ation of them will suffice .
As regards temperature of the pendulum , it will be sufficient to assume that the whole expands uniformly with rise of temperature .
Let coefficient 'of linear expansion at , then the time of swing at will be , where Thus for a rise of one degree we reduced the observed result by , where is the time at the standard temperature With to the effect of changes in the density of the air , we need only consider buoyancy and the mass of air carried along with the pendulum .
This latter consists of two parts , namely , that which surrounds the pendulum :and moves with it , and that within the hollow bob ( which is not air-tight ) between the particles of the substance therein contained .
Some air will pass in and out of the bob as the conditions change .
We shall calculate the buoyancy as affecting the whole volume of the bob , as well as the rest of the pendulum , and therefore any air which enters the bob must be considered to add its weight to the bob as well as its mass .
We thus have to apply corrections for\mdash ; ( 1 ) A variation in the weight of the moving system , equal to the variation in the weight of the air displaced by the whole pendulum , and acting at the centre of gravity of this displaced air .
( 2 ) A variation in the moment of inertia of the system , equal to that in the air carried along around the pendulum .
( 3 ) A variation in the weight of the bob , equal to the variation in the weight of the enclo air .
.(4 ) A variation in the moment of inertia of the bob , equal to the variation in that of the enclosed air .
1910 .
] Mass to for Radioactire Substance .
331 The standard invar pendulum is also subject to corrections ( 1 ) and ( 2 ) .
The effect of this is to partially eliminate the corrections , and in view of this it will be allowable to make the assulnption that the mass of air carried with each pendulum is equal to the mass displaced .
to corrections ( 3 ) and ( 4 ) , will be assumed that the density of the air inside the bob is equal to that outside .
The corrections will be considered in detail in a later section of the paper .
3 .
of th The main apparatus is shoWll generally in fig. 1 , and an portion in profile in fig. 2 .
The same letters are used for corresponding parts in the two figures .
, A represents the rod of the experimental pendulum , is the bob , the cast-iron supporting bracket , which is slotted at and which carries slotted plates , , which can be levelled , and to which are fixed planes , , on which the of the pendulum bear .
The upper plate is thrown back 1 cm .
, as shown in fig. 2 .
The pelldulum is there shown with the upper knife-edge in contact with its planes , and in this position the lower is free .
On the other hand , when the lower knife-edge is in position the upper one ings clear of its planes .
This change of position is accomplished by means of the apparatus in the ures .
Two pairs of brackets , , are attached to a flat rod , this lies exactly behind the pendulum rod in fig. 1 ) , which is moved its own plane by eccentrics , , so that any point in it describes a part of a circle whose radius is 5 mms .
This throw is slightly erated in fig. 2 for the sake of clearness .
The arms , , have -shaped grooves , in which pins , are seen lying in fig. 2 .
These pins are fixed to the pendulum , and by means of them the are , lift the pendulum and carr ) it over from ] position shown in fig. 2 to that in which the lower knife-edge engao e with its The arms then continue their motion htly , so as to release the pins , and allow the pendudum to clear of its supports .
The eccentrics actuated by rods , cranks , and connecting rod , as shown at , in fig. 1 .
These pass outside the case in which the pendulum is enclosed , so that the change from one knife-edge to the other can be elf'ected without the case .
This is an important point in experiments of this nature .
A mirror is fixed to the pendulum at , and another one , , which emains stationary during an observation , is attached to an adjustable support Q. This has two motions controlled rods which pass the case at .
The bracket carrying this fixed mirror arrangement is at S. A )racket with two screws is shown at , near the bob in .
The screws are adjusted so as to very nearly touch the rod whenever the bob is for 332 Mr. L. Southerns .
FIG. 1 .
FIG. 2 .
: 1910 .
Mass to Weight for a Substance .
, 333 : ' replaced .
They are for the purpose of steadying the rod , and are screwed out of the way when the bob is in position .
is a rod similar to the pendulum rod .
It has a brass projection , which is drilled to receive the bulb of a thermometer W. The reading of this thermometer is taken as indicating with sufficient accuracy the temperature of the pendulum rod .
A thermometer is also suspended in the case to indicate the temperature of the air .
The standard invariable pendulum , to which all observations are finally referred , is shown to the left of the experimental pendulum , in fig. 1 .
The rod is of invar , and the bob of brass .
It is provided with mirrors , similar to those of the other pendulum .
The pendulums are set up in a niche in the wall of one of the rooms ou the ground floor of the new of the Cavendish Laboratory .
The cast-iron brackets are bolted to the wall which forms the back of the niche .
Double partitions of wood , packed with sawdust , are fixed above and to the right of the pendulums , and the front is closed with two doors similarly made , an upper and a lower door .
The lower one can be opened independently , when it is necessary to remove and replace the bob .
A plate glass window is fixed in the upper door opposite the mirrors .
This is closed by a shutter , except when observations are being made .
There is also a partition between the two pendulums .
No culty was anticipated , nor has any been experienced , due to the fact that both pendulums are fixed to the same wall .
Their periods are quite different , and even if any slight resonance did occur , it would not favour either of the substances llsed .
( There is , moreover , no necessity taking observations on both pendulums simultaneously .
It is sufficient , if a good clock be used , to take them alternately .
) To set the pendulums in motion , tubes with rubber bulbs are used , by means of which puffs of air can be projected against the bobs .
In taking observations a flashing apparatus is employed .
It consists of a clock capable of giving electrical signals each second , a lever actuated by these and arranged so as to interrupt the primary current of an induction coil , the said coil , and a helium tube , which is connected to the secondary of the coil .
For most of this apparatus , the writer is indebted to Dr. F. Horton , who has described it in detail .
* The helium tube is set up horizontally on a shelf fixed to the wall of the room opposite the pendulums , at a distance frolll them of metres and about metre above the level of the mirrors .
cylindrical concave mirror is placed slightly behind and below the tube , so as to give a fine and sharp line-image each time a flash occurs .
This imsge is observed by reflection in the pendulum mirrors by means of telescopes , one * See ' Phil. Trans , vol. 204 .
VOL. LXXXIV.\mdash ; A. 2 about 1 metre in front of the mirrors .
The telescopes tilt slightly upward towards the mirrors , and the rays from the flashing apparatus to the mirrors pass just above the observer 's head .
As the clock which controls the flashes is in another part of .
the laboratory , a chronometer is placed near the telescopes for the indication of time .
4 .
Method of Experiment .
It is necessary , as has been explained in section 2 , to make determinations of the of swing of the experimental pendulum about its two knife-edges for various positions of the bob , one of the substances to be experimented upon being con tained therein , and then to change to the second substance and repeat the series of observations .
As , however , only comparative determinations are required , it is not necessary to express the times in seconds .
Any other invariable unit of time will do equally well .
The unit actually adopted is nearly 1 second , it is , where is the time of a complete swing of the standard pendulum under normal conditions of temperature , air density , and amplitude of oscillation , and is approximately the average value of the time of the standard as determined directly in the experiment in clock seconds .
is assumed to be accurately constant during a series of observations .
Its absolute value is not required , and has not been determined by reference to transit is , of course , very nearly .
Corrections are made for changes from the normal density of the air , but corrections for variation of amplitude and temperature of the pendulum itself ( the rod of which is of invar ) are negligibly small and have not been applied .
Now suppose the time of the experimental pendulum , and also that of the standard , to be observed in clock seconds , then if their actual times are and in mean solar seconds , and if the clock second is equal to mean solar seconds , the observed times will be and respectively , so that we shall have , as the ratio of the observed time of the experimental pendulum to that of the standard , , which is the actual time of the experimental pendulum in terms of that of the standard .
In order to express this in lJnits more nearly equal to 1 second , may be multiplied by which will give the actual time of the experimental pendulum in terms of , the unit described above .
The finally tabulated times of the experimental pendulum will be given by\mdash ; time bservations ancidences Prof. Pointing , and described by Dr. Horton in the paper referred to above .
One or two modifications in the procedure have however been made in the present case .
The method as used will be understood the following description .
Suppose the pendulum which is to be timed the flashin apparatus to be working , we shall observe , on looking through the telescope , whenever a flash occurs , two images of it in the field of view .
One of these , due to the fixed mirror , will have a constant positio1i in the field , the other , due to the mirror attached to the pendulum , will in general be seen in a new position each second .
The fixed mirror should be adjusted that when pendulum is vertioal , the two images are seen to be in the same line and partially overlapping .
Now suppose that at some instant a perfect coincidence of the images occurs .
If there is no special relation between the period of the pendulum and that of the flashes , the moving image will then appear to dart about the field in an erratic manner .
After a time it will occupy a position near to the fixed image , but probably not exactly coinciding with it .
Let this occur seconds after the first coincidence .
At this instant the pendulum be nearly vertical , and if it is moving in the same direction as when the first coincidence occurred it will have executed almost , but not quite , exactly a whole number , say , of swings .
Suppose it to have passed a little beyond its central position , so that the actual numbel of swings is , where is a small fraction .
In this case the pendulum may be said to be gaining on the flashes .
In seconds the pendulum will have executed Again , in another seconds the moving image will appear a little further from the zero ( or fixed ) , and so on .
After a time the image will reach its maximum distance from the zero , and at that installt the pendulum will be at one extremity of its swing , and it will have ained one-quarter of a complete period on the flashes .
The image will then begin to return towards the zero .
When it reaches it , the pendulum will have gained half a complete .
The image will then pass along to the other extremity of its range , and again eturn to the zero .
We will suppose that an exact coincidence then occurs .
At that in stant the pendulum will have ained exactly one complete swing on the flashes , and the observation will terminate .
All the flashes except those occurring at intervals of seconds will , of course , have been ignored .
Let the total number of periods of seconds each be , then we know that the pendulum will have executed swings in seconds , and its time of swing is thus .
In general , however , exact coincidences do not occur , and it is necessary to observe two successive positions of the image , one on each side of the zero , the observation , should a perfect coincidence not occur .
In order to enable these distances to be estimated , scales , slightly illuminated , are used in the eye-pieces of the telescopes .
There are two sources of error in practice which need attention .
In the first place , an error may arise due to an irregularity of the flashes caused in the following way .
The clock pendulum is fitted with a platinum projection , which cuts through a fine stream of mercury and so makes contact each time it passes its mid-position .
Now , should the projection or the mercury stream be slightly displaced laterally , so that contact always occurs when the pendulum is a little to one side of its mid-position , it is evident that the contacts will take place at unequal intervals .
Alternate contacts will , however , take place at two-second intervals .
Thus in the actual observations must always be an even number ; this ensures freedom from error due to the above cause .
Again , the images in the field of view may not exactly coincide when the pendulum under observation is in its central position , and in this case the true zero is not accurately indicated by the fixed image .
would be of little importance if the amplitude of the pendulum remained constant .
But , as this is not the case , an error is introduced if we take the fixed image as representing the zero .
This error is practically eliminated by two independent ( but overlapping ) observations of the time , one being started with a coincidence which occurs while the pendulum is moving towards the observer , and the other with one which occurs when the pendulum is in opposite phase .
The mean of these times is to be taken as the observed time .
The actual procedure is as follows :First , the experimental pendulum is set on its lower knife-edge and set into motion ; the standard is also started .
After 10 or 15 minutes the two observations on the experimental pendulum are commenced , and then the two on the standard .
In due course the experimental observations are completed , the pendulum stopped , and set on its upper and restarted .
After a few minutes two observations on it commenced ; these are completed , and finally the two standard observations are completed .
The value of for each of the six observations is then recorded- ecords are also kept of initial and final amplitudes , of the readings of the thermometers and of the barometer , ' and of a maximum and minimum thermometer ( in the mornings only ) kept in the case to show the range of temperature during the previous 24 hours .
1910 .
] Mass to Weight for a 0037 The filling of the bob is an important operation .
As has been shown , slight variations in the position of the centre of gravity are in themselves unimportant , but errors are introduced by changes in the nloment of inertia - of the bob about an axis its centre of gravity .
With reasonable care these can be rendered small .
The substances used are red lead and uranium oxide .
To the latter , however , it has been found necessary to add a small proportion of nagnesium oxide , to render it similar to the red lead in packing properties .
The method of packing finally adopted is as follows : substance is divided into three equal portions by weight , and one of these is put into the bob .
The bob is then tapped underneath , to cause the substance to nearly settle down into the lower third of the bob .
Then a flat disc of brass is lowered on to the substance , and on this is placed a distance piece , which is pressed down until its top is exactly level with the rim of the bob .
By this means the substance is htly compressed into a definite volume into the lower part of the bob .
The disc , etc. , are then removed and the second poltion of the substance is put into the bob , and a similar process oone through , .
a shorter distance piece .
Finally , the remainder of the substance is put in and tapped down as before , and pressed home by means of the lid of the bob , which is then secured by its nut , care being taken to always screw down this llut to the same extent .
Greater accuracy of packing could be attained by the stages of the process , but the three have been found to be sufficient for the purposes of this experiment .
As regards the amplitudes of oscillation , these were practically identical in the several experiments .
The initial and final amplitudes were\mdash ; Standard pendulum Exp. pendulum , upper K.E. lower K.E. 5 .
Experim Results .
In Table I some constants of the pendulum are given .
They have been carefully determined , *though strict accuracy is not essential in the case of quantities which lemain constant throughout the whole series of experiments .
Table II summarises the records of observations of the final experiments .
Two experiments were usually made in a day , one in the morning and one in the afternoon .
The substance was usually changed once day after the morning orvation.table , alues fthese constants Ixperiments taries , espect tumbers in the table actually cosition osubstance hssumed i 338 Mr. L. Southerns .
Determi ' 910 .
] Mass to Weight for a Radioactive Substance .
S39 are given for the standard and for the upper and lower knife-edge observations on the experimental pendulum .
The values of and , and the number of swings gained or lost during the observations , are as follows:\mdash ; Standard pendulumains 1 swing . .
pendulum , upper ains 1 swing .
lower K.E. : , loses 2 The position of the bob is given in each case , ndicates n position , and a position somewhat lower than the normal lnd different in the different cases .
The other columns of the table require no special explanation .
The observed times , as obtained from the values of by the method which has previously been explained , are given in Table .
Three sets of corrections are also given in the table .
has been shown , it is only necessary to correct for variations from the normal conditions under which the experiments are made .
The normal temperature is taken to be C. and the pressure .
In the first place , corrections given for deviations in the temperature of the experimental pendululn from These are obtained in a manner which has been described in section The coefficient of expansion is taken to be .
Taking approximate values of for upper and lower knife-edges , we have for upper knife-edge sec. for lower , , , , Thus for a rise of 1o C. we reduce the times by the above quantities .
Correctione are next given for variations in the pressure .
In order to obtain these it is necessary to know the increments of the values of I and in the expression for the time of swing of the pendulum under consideration .
Thus , for example , in the case of the experimental pendulum swinging about its upper knife-edge , we have say .
From Table I we see that Now , from a knowledge of the dimensions of the apparatus , it is easy to calculate with sufficient accuracy the produced in I and by an increase of pressure of 1 cm .
of mercury .
In making this calculation , regard must be paid to the points mentioned on p. 330 .
In the above case , an increase of pressure of 1 cm .
gives rise to an increase of I of 167 , and a decrease of of .
Thus the new time will be given by , so that we have of tloe Ratio of \mdash ; $ 1910 .
] Mass to for a which is found to mean that the increase of pressure causes an increase of represented by units in the last place of decimals tabulated .
Therefore , to reduce the observed times to times for pressure , it is necessary to deduct units from the last } place for each eentimetre of pressure .
greater than 76 .
In the case of the knifeunits are to be deducted , and in the case of the standard , 4 units .
The effect of a fall of 1o C. in the temperature of the air is to that of an increase of pressure of 1 cm .
as 76 : 285 .
The ection s given in ) table for changes of atmospheric temperature are based on this fact .
Table The corrected times are given in Table , together with the " " tabulated times In these , as was explained on p. 334 , the clock errol .
S are elinlinated , and the times of the experimental pendulum are expressed iu invariable units closely approximating to seconds .
The tabulated times can now be plotted , and the characteristic lines obtained for the two substances .
We may first , however , determine the amount by which they will be separated on the supposition that the theory given at the of the paper holds true .
We first require the approximate mass of radium equivalent in the radioactive substance used .
The mass of uranium oxide is 1015 grammes , and assuming the composition to be given by , we shall have 860 gramrnes of urimium , corresponding to 811 grammes of radium .
The extra mass attached to each ramme of radium is ramme .
This gives 62 milligrammes for 806 rammes * Week ending Apri116 .
Week ending Apri ] 1910 .
] Mass to Weight for Substance .
343 6 .
It is evident from the diagrams that the characteristic lines for the two substances are identical within the limits of experimental error .
In the experiments exhibited on the first diagram , the accurate method of packing ' was not in use , and the weather conditions were not quite so favourable as they were .
the week following .
In the second oram , the of the better system of packing , and also of the more uniform weather which prevailed , are well shown .
The dotted lines are drawn ] to the experimental characteristic lines , so as to show the distance which should separate the lead and uranium lines , according to the theory .
In the more accurate , the deviations from the line are not greater than 1 in 600,000 on the times .
The results indicate that the ratio of mass to weight for uranium oxide does not differ from that for lead oxide by more than 1 part in 200,000 .
It remains to show that inaccuracies in the crhing of the bob are not likely to cause appreciable errors .
Following the method of p. 339 , we have for the normal time from the upper Now , suppose we add gramme to the bob , the increase in I ( the numerator ) will be about 196 , and the increase in Mh ( the denominator ) about .
Thus , for the new time , we have which gives or , which is an increase of 1 in 2,000,000 , and is therefore negligible , which is also the case if the lower knife-edge be considered , though the effect is somewhat greater , , about 1 in 300,000 .
It is , of course , easy to reduce the errors to smaller dimensions than these by careful weighing .
An effect which caused considerable trouble the experiments may be briefly referred to .
A rod of nickel steel was first used on account of its small coefficient of expansion .
Sometimes for two or three days satisfactory results would be obtained , then the characteristic line would suddenly take up a new position to the left of the old one .
This took place time after time , the line always moving parallel to itself in the same direction .
A rod of tool steel exhibited the same error .
It appeared to be due to magnetic effects .
The rods became magnetised by induction in the earth 's field , and no doubt this magnetisation was assisted by the movements of the pendulum when it was transferred from one knife-edge to the other , and the removal and readjustment of the bob .
The cast-iron bracket was probably concerned in the matter .
The rod of the standard pendulum was also magnetic , but it remained untouched during the whole course of the experiments , and gave rise to no difficulty .
It had , at any rate , settled down to a The pendulum was sensitive to sudden variations of the weather .
It would be a great advantage to enclose the apparatus in a case kept at constant temperature and pressure .
A somewhat improved method of removing and replacing the bob is also desirable .
In conclusion , the writer wishes to express his best thanks to Prof. Sir J. J. Thomson for his unfailing kindness throughout the course of this investigation .
On the Spontaneous Crystallisation the Melting Freezing Point Curves of JMixtures of Two Substances which form Mixed possess Minimum or Eutectic Freezing Point.\mdash ; IVIixtures of Azobenzene and Benzylaniline .
By Miss F. ISAAC , formerly Research Fellow of Somerville College .
( Communicated by Principal H. A. Miers , F.R.S. Received June 21 , \mdash ; Read June 30 , 1910 .
) CONTEN TS .
PAGK Crystalline Form of Benzylaniline and Azobenzene 345 Microscopic Examination of Azobenzene and Benzylaniline crystallised from Fusion 346 Mixtures of Azobenzene and Benzylaniline 347 The Freezing Point Curve 348 The Melting Point Curve 348 Analysis of the Mixed Crystals Microscopic Examination of Mixtures of Azobenzene and Benzylaniline crystallised from Fusion The Supersolubility Curve , or Curve of Spontaneous Crystallisation 355 ( 1 ) By the Method of Sealed Tubes ( 2 ) From Observations of the Refractive Indices of Liquid Mixtures 367 Nature of the Crystals separating on the Supersolubility Curve 360 Thin Sections cut from Solidified Mixtures of Azobenzene and Benzylaniline 366 Conclusion The behaviour of mixtures of naphthalens and -naphthol has already been investigated , and the freezing melting point curves and the curve of

