286 Prof. W. J. Sollas .
scratch the face of the crystal .
It was found that if it requi ] f load of one to produce scratching parallel to the edge of a face o ; he cube , a load of 13 was necessary parallel to a diagonal .
Look broadly , it will be seen by reference to the figure that the effectf a \gt ; force acting parallel to a diagonal should be to that acting pa ?
lei 14 to a side , as 1 : \/ 2 , i.e. , as 1 : P414 .
Considering how compliced fthe problem actually is , this correspondence is quite as close as old # possibly be expected .
By several distinct lines of argument , resting on the stud of f\lt ; molecular volumes , of the density of solutions , of the symmetry nd a physical characters of crystals , we have been led to the same resit ; d given the atomic volumes in the ratios we have assigned to thermo other was possible ; and we may now with greater confid ce .
3 proceed to the investigation of other diatomic compounds , w will be found to throw a surprising light on the molecular tactii of f crystals .
" On the Intimate Structure of Crystals .
Part II .
Crystal ihe Cubic System with Cubic Cleavage .
Haloid C pounds of Silver .
" By W. J. Sollas , LL. D. , D.Sc .
, F.b Professor of Geology in the University of Oxford , o ceived January 27 , \#151 ; Read February 3 , 1898 .
The haloid compounds of silver offer more points of interest in study of crystal tactics than those of the alkalis , especially as by include that remarkably anomalous substance\#151 ; iodide of silver .
Silver itself crystallises in the cubic system , and possesses same kind of structure as all other metals which possess c symmetry , i.e. , its atoms are arranged on the plan of most c packing : unlike the alkali metals it undergoes no change of voir on entering into combination .
Its atomic weight is 107'93 ; density has been differently determined by different observers , silver heated in vacuo Dumas found 10'512 , and Roberts-Am 10'57 ; we take the mean of these numbers 10'541 .
The atoi weight , 107'938 , divided by 10-541 , gives 1C233 , the gross atoi !
volume .
Prom this the diameter of the atomic sphere is found to 2T7152 , and its volume 5'3616 .
Both the chloride and bromide crystallise in the cubic syst .
and are regarded from analogy as possessing the same crystal structure and symmetry as the corresponding salts of potassium r sodium .
The specific gravity of silver chloride is obtained from the scatter fusion .
The latest determinations give for silver chloride On the Intimate Structure of Crystals .
; pific gravity of from 5*517 to 5*594 , and for silver bromide from \lt ; \gt ; .\#166 ; 5 to 6*425 .
Data for calculation are given in the table :\#151 ; .g . . .
lg01 .
igBr .
-gl-2 \#166 ; M. W. Specific gravity .
Gross volume .
Diameter of atomic sphere or edge of crystal cubelet .
107*938 10 -541 10 -233 2 -1715 143 *391 5-517 26 *249 4 -7175 ( a ) 5*594 26 -615 4-7395 ( J ) 187 *901 6-215 30 233 4-9452 ( a ) 6-425 29 -246 4-8907 ( 6 ) 234*798 5 675 41 -374 Salt , gross volume .
Silver , gross volume .
Difference of gross volume .
AgCl ... .
26 -249 10 -213 16 *036 chlorine ( a ) 26-615 If 16*402 " ( b ) AgBr. .
..29-246 99 19 *033 bromine ( a ) 30 -233 99 20 *02 " ( b ) Agl 41 -374 99 31*141 iodine .
Ve are now in a position to compare the diameter of atoms of v er in the free state and in combination .
The diameter of chlorine og taken as 2*4954 , the value found in Part I , and that of bmine 2*801 , its value in sodium bromide , we obtain from Prohor Miers ' formula\#151 ; Diameter of Ag in\#151 ; * AgCl .
AgBr .
( a ) 2*2279 ( a ) 2*1784 ( b ) 2*2399 2*1307 Phe results for ( a ) and ( b ) correspond to the different specific i-.vities ( a ) and ( b ) given above , and obtained by different Servers .
Phe differences between the diameters ( a ) of silver in combination ; il that it possesses in the free state are not great .
In comparing the gross volumes of the haloids given above with r ir gross volumes in the haloid compounds of the alkalis , it will ^observed that the gross volume of bromine in silver and sodium Wides is closely correspondent ; this is because the atomic volumes ^sodium and silver have nearly the same relative value .
Silver iodide , to which we now turn , is one of the most interesting ^stances we have yet encountered , and most important results flow from its study .
At ordinary temperatures it does not present y 2 Prof. W. J. Sollas .
itself in the cubic system , but in hemimorphic hexagonal cry , ] .
For this reason we cannot , by the method we have previously ployed , calculate the diameters of its component atoms , but a rer .
ence to the preceding table will show that the gross volume o'he iodine is not greatly in excess of that which it possesses in soma iodide , so that in all probability its atomic volume is the sam M sufficiently similar to serve as basis for argument .
The diamet of the silver atom is then 2d 715 , that of the iodine 3T905 .
The tio of the diameter of the silver to that of the iodine falls conseque lie below the limiting rates of 07286 : l'O .
The exact ratio bot or this and the constituents of the other silver compounds is give in the following list:\#151 ; Ratio of diameters of atoms .
Ag : Cl = 08928 : 1 Ag : Br = 07761 : 1 Ag : I = 06894 : 1 From this it is evident that while the chloride and bromide are i ' within the 0"728 limit , the iodide lies outside it ; and it follows 1 our hypothesis that while the two former salts may be , as they cubic , the iodide should not , but might be hexagonal hemimorp as it is .
When silver iodide is heated it contracts along the vertical axi and expands in directions normal to this , till at a temperatui i 146 ' C. it passes per saltum into the cubic system , and thenv ' further rise of temperature expands uniformly .
The transi from the cubic to the hexagonal system on cooling is accompli^1 almost with explosive violence , fragments of the solid iodide !
projected into the air , and deep clefts extend into its substa How on the basis of our reasoning is this curious transformatio be explained P Only on the supposition that the atomic volu expand as the temperature of the substance rises ; the mass , t-1 whole , may contract , as it certainly does , but this is to be accoui * for by a change in the relative position of the atoms , the ato volumes themselves must be conceived as all the time expandi .
But it may be asked , is there any reason why the iodine and sil should expand at the same rate , and if not , which is the more lit ' to possess a higher coefficient of expansion , the iodine or the sil atom ?
The probabilities are all in favour of the iodine .
The^j efficient of expansion of silver in the free state is 0-000037 to the melting point , and 0"00002 from 0 ' to 100 ' C. That iocf has an immensely greater coefficient may be judged from the char-in its specific gravity with rise of temperature , thus at its sp. gr. is 4-917 ; at 60 ' , it is 4*886 ; at 79-6 ' , it is 4-857 ; at 1 ' just before it liquefies , it is 4"825 ; after liquefaction , it is 4 On the Intimate Structure of Crystals .
289 - it 151 ' , a point just above the temperature at which the hex-riil form of silver iodide is exchanged for the cubic , it is 3'866 .
may fairly therefore attribute the greater expansion to the , Mle , the silver probably undergoing but slight increase in bulk .
n hat way does this help us ?
If the disparity between the .
lues of the atoms were originally the cause which determined besilver iodide to assume hexagonal instead of cubic symmetry , \ J by increasing this disparity , shall we render it cubic ?
Nothing i !
ioe simpler .
By the time the substance has reached 146 ' C. the f aic volume of the iodine has become so great relatively to the \#187 ; hr that the latter is lost in the interstices , and it is the iodine i'os which is directly operative in supporting the crystalline iae .
Here we have a basis for calculation .
If the iodine thus , hs the framework of the crystal , it must be , as we shall show ' because it is packed on that open system which we have \gt ; lady described and illustrated in the first part of this communion , From this the diameter of the iodine spheres can be calcu-latl .
ae volume of silver iodide , as deduced from Rodwell 's data , is U'7 , just above 142 ' , the point which Rodwell gives as the critical leiperature for the change of system .
Supposing , as we have indy stipulated , that the iodine is built up in most open cubic anting , we find from this volume the number 3'439 as the length :i ie diameter of the iodine atom .
re shall make important use of this number directly , but we nt first endeavour to find a probable structure for the silver iodide w n crystallised in the hexagonal system .
; will conduce to clearness to avoid discussttyi at this stage , and to escribe at once the structure which I am led to think is the only fiqible one consistent with the crystalline and physical properties the compound .
Let us regard the silver and iodine spheres as viietly united together to form molecules of Agl ; conceive these led on a horizontal plane with their major axes vertical , and disced in triangular order as in the diagram ( fig. 1 ) .
The balls must i * be in contact laterally , but separated by definite regular intervals ; ovthis first sheet place a second similarly formed , and so related to ill'first that the silver atoms rest in the alternate intervals left between h lower set of iodine balls taken in threes ( fig. 2 ) .
The assemblage Produced will be homogeneous , hexagonal , and hemimorphic .
It Resents silver iodide as it exists at ordinary temperatures .
Now , luring in mind the fact that every silver atom rests in the interval thween three iodine atoms , let the latter enlarge to a slight extent , ii84 simultaneously retreat from each other along the horizontal ?
Pne to a small but uniform amount .
This corresponds to the Mansion along the lateral axes on heating .
But as a result of this Prof. W. J.'-Sollas .
i Tie .
1 .
Plan of first sheet of molecules of Agl .
One molecule of Agl shown in elevation .
Pig .
2 .
Plan showing two sheets of molecules of Agl .
On the Intimate Structure of .
291 insion the atoms of silver will descend deeper and deeper , / even the atoms of iodine , and the distance between successive \gt ; ts in a vertical direction will diminish .
This corresponds to raction along the vertical axes with rise of temperature .
On hypothesis the anomalous contraction of silver iodide ceases to , ,n anomaly , but follows as a natural , though by no means neces-consequence of its crystalline structure , t the critical point , when the hexagonal is exchanged for ic symmetry , the atoms of iodine possess , as already shown , a i neter of 3-439 .
Let them be represented in plan in the relative itions they must occupy for open cubical packing .
They are wn in fig. 4 by a section taken at right angles to a trigonal iA .
leasured along the edge of the cube , the intervals between the ns from centre to centre must , on this system of packing , be to diameters precisely as v72 : 1 .
Find the diameter of a small ere that will just fit in between the three spheres as represented he section .
It is exactly 2*17687 .
But we have already found tit the diameter of an atom of silver , both in the free state and in \#166 ; ibination , measures at ordinary temperatures 2 1715 , an extra-\gt ; inary coincidence .
The difference between this number and * t just obtained , 2T769\#151 ; 2T715 = 0'0054 ; again , the difference -ween the diameter of the atoms of iodine in the alkaline iodides irdinary temperatures and the atoms of iodine in silver iodide at ; ' is 3-439\#151 ; 3-2 = 0'239 , and these two numbers ( 0-0054 and 39 ) probably approximately represent the change in dimension he constituent atoms of silver iodide as the temperature of this - tstance is raised through a range of 130 ' .
Calculating from the fficients given by Fizeau , the expansion of 0"0054 , which we have nd for the silver , is just twice that which it would experience in free state .
Jp to the critical point , when the interval between the iodine ms has become large enough to allow the atoms of silvei to pass ween them , the atoms of silver have been supported by the atoms iodine and have supported iodine in their turn ; all the atoms ve been subject to molecular pressure , but directly the critical lsition is attained , the atoms of silver are driven by this internal assure into the middle of cubical clusters of atoms of iodine and rtly or wholly released from pressure , the atoms of iodine alone 3n sustaining the crystalline fabric .
It is not wonderful un^ei \gt ; es circumstances that the spectrum of silver iodide should differ irkedly when observed below and above the critical point , thus ernicke has observed that at a high temperatuie the spectrum ( o 'ver iodide ) , like that of solid and liquid iodine , contains no blue r violet light .
In the normal state , below 138 ' C. , silver iodide Prof. W. J. Sollas .
gives a spectrum less bright , but twice as long and particular developed in the blue and violet spectrum.* The passage of silver iodide from one form to the other is as as been stated , sudden and abrupt .
This necessarily follows , fromie geometrical conditions of the case .
On reference to fig. 5 it wily , seen by inspection that while a gradual descent of the atomsaf silver may take place so long as their centres are situated above centres of the atoms of iodine , against which they glide , yet direiy after they come to lie in the same plane , a sudden descent must fee place to a definite extent , which is given by the formula DO !
\#151 ; 1 ) ?
= \#187 ; " i+ ?
2\#151 ; 2ri cos 0 , or by the equivalent \#151 ; 4 where ris the radius of the large spheres , ra of the small spheres , and 0i02 e distance through which a sudden descent occurs .
The instantaneous descent of the silver which thus takes placs accompanied by a sudden change of volume in the compound itse ; as determined by Rod well , the volume diminishes from l,01575o l'O , as the temperature passes above 142 ' , the maximum density !
the salt then being attained .
In the collocation of spheres win we have imagined , it is possible to compare the bulk before ai after the critical point is passed very simply .
A tetrahedron ' constructed by joining the centres of three spheres of iodine belt , which are in contact with one of silver lying on the axis above , w\#171 ; a fourth of iodine , also on the vertical axis and attached to o single sphere of silver .
The ratio of the volume of this tetrahedn is directly proportionate to the whole volume of the sfcructu , whether in the cubic or hexagonal systems , and at the critical po the volumes are directly proportionate to the heights of the respt tive tetrahedra .
From this we find that the volume before contract !
is to that after as 114 : 100 , amply sufficient , and , it might objected , superfluous ; for the contraction , as observed by Rod we only amounted to from 1016 to 1000 .
This is a case , however , which theory proves more correct than observation , for Mallard a : Chatelier have shownj ' that Rodwell 's results are erroneous , owii as they remark , to his having deduced the cubical expansion fro the linear extension , as though silver iodide were an isotropic bod These observers were able to bring about the transformation from t-hexagonal to the cubic structure by the application of pressui ( 4,000 kilograms to the square centimetre ) , and they found tb ; the ratio of the volumes before and after change , was as 116 to 10 which gives a coefficient ten times as great as that of Rodwell , an very closely in agreement with that ( 114 ) which we have theoretical * Wernicke , ' Pogg .
Ann. , ' vol. 143 , vide Rodwell , 'Proc .
Boy .
Soc. , 'vol .
2 p. 206 .
Wernicke accounts for this by supposing some of the iodine to be liberat* from combination .
f 'Bull Soc. Min .
, ' vol. 7 , 1884 ; ' Journ. Phys. , ' vol. 4 , p. 305 , 1885 .
On the Intimate Structure of Crystals .
duced .
Under the influence of rising temperature , the change smaller , from 111 to 100 , but this value is said to be merely oximate .
That the change of volume should be smaller is not ^intelligible ; the direct effect of pressure is to bring about a ange of configuration , which is unstable , since the substance i/ erts to its original state directly the pressure is relieved ; the eect effect of heat , on the other hand , is to produce an expansion ( the atomic volumes , and the change of configuration follows only ; a consequence of this .
In the difference between the value 111 \#166 ; d 116 we should have a measure of this increase in atomic lume , but for the probability that an expansion of atomic volume ay take place , as a consequence of the change of configuration lich results from the action of pressure .
It is of interest to note at the amount by which the pressure must be reduced to reverse e operation and bring back the substance to the original state is ily one-half of that required to produce the change directly ; the versal also takes place much more rapidly than the direct trans-rmation .
It is to the beautiful observations of Fizeau that we owe our lowledge of the change in .
dimensions of silver iodide measured in lation to its crystalline axes .
Along the axis c , Fizeau found a jgative coefficient of expansion of -0-00000397 ; along two rect-igular axes in a plane normal to c he found a positive coefficient of 00000065 .
These were for a mean of 40 ' over an interval of from -10 ' to 70 ' ; the coefficient of variation Aa/ A0 was found to be for -4-27 , for the other axes 1-38 .
Thus for the mean temperature le coefficient of contraction is six times that of the coefficient of mansion .
A very significant relation , but not more so than the ict that the contraction increases as the temperature rises , and , as indwell 's observations seem to prove , the increase becomes very onsiderable as the temperature rises above 70 ' .
Every geometer rill perceive at once that these relations are m a so n e iony with the conception we have framed of the ordering of the mecules in the crystal .
They directly depend on the changing atio of the sine and cosine of the angle 6 as the small sphere f the figure ( fig. 6 ) is squeezed out from the three larger spheres I)against which it is pressed ( only one of these spheres ( ) i hown in the figure ) .
As the iodine atoms approac eac o , nth fall of temperature , the line aO will revolve roun a as a entre , and thus the vertical parameter will expan as ontal contracts .
From the coefficients given hy Fizeau , it fo hat at his mean temperature of 40 ' the angle is \gt ; , , , irystallographic measurements of Zepharovich , ma e presum he ordinary temperature of the air , and therefore a oi a his angle appears to have become 22 ' 17 ' , as calculated for me by 294 Prof. W. J Sollas .
Fig. 3 .
Vertical section through one edge and centre of a face of a rhombohedron of three sheets .
Fm .
4 .
Fig. 5 On the Intimate Structure of Crystals .
295 .ofessor Miers .
The ratio of sine and cosine for this angle is ill : 1*0 , which gives the relation that should subsist by hypothesis tween the coefficient of expansion along the axes a and c at 0 C. Fizeau 's observations fail ns here , and I look forward th interest to fresh determinations of these constants , which r. Tutton has kindly promised to make for me .
It is now ssible to trace out the series of changes which silver iodide un-rgoes as it falls in temperature from 150 ' downwards ; at 146 ' lallard and Chatelier ) or 142 ' ( Rodwell ) it possesses cubic symmetry , cause the atoms of iodine are out of all proportion so much rger than those of silver , that the latter cease to be operative as aments in the crystalline structure ; the iodine atoms are arranged form cubes which we may picture to ourselves as rhombohedra ith an angle of 90 ' ; every iodine atom is chemically united ith an atom of silver which depends from it into the central space ithin a rhombohedron ; as the iodine atoms become smaller , with ss of energy , this space becomes too small to contain the silver om which is consequently forced outwards or upwards ; as it passes it of the rhombohedron of 90 ' the vertical parameter of the latter necessarily lengthened , while its horizontal parameters are lortened ; the change in dimension of the vertical parameter , hich occurs as soon as the centre of the atom of silver begins to Love above the plane containing the centres of three surrounding idine atoms , is excessive as compared with that of the horizontal arameter , but the difference is diminished by the diminishing ratio etween the diameters of the two different kinds of atoms ; at 15 the ifference of the two coefficients is probably nearly as 3 :1 ; it will dimi-ish rapidly with further cooling , and , before thqatoms of silver assume position in which the radius vector drawn from them to the iodine toms , about which they revolve , makes an angle of 45 with the arameter Oa , the expansion along the vertical axis will cease and e replaced by contraction , unless , indeed , as this crisis is approached he diameters of the atoms should fall within the ratio of 1 : 0 729 , ffien the crystal would once again become cubic .
In the study of molecular volumes we hold the key which is estined to unlock the secret of crystal stimcture so long concealed .
Ve shall eventually discover by its means , in connection with other tudies , the relative and absolute dimensions of all the elementary toms , and probably not only this but also their true forms\#151 ; foi it s possible that they are not all spheres .

