]\gt ; Progressive lightning channel is ascribed to the existence of a second centre of charge within the cloud .
REFERENCES Schonland , B. F. J. 1937 Phil. Mag. 23 , 503 .
\mdash ; 1938 Proc. Roy .
Soc. A , 164 , 132 .
Schonland , B. F. J. , Hodges , D. B. and Collens , H. 1938 Proc. Roy .
Soc. A , 166 , 56 .
Schonland , B. F. J. , Malan , D. J. and Collens , H. 1935 Proc. Roy .
Soc. A , 152 , 595 .
Workman , E. J. , Beams , J. W. and Snoddy , L. B. 1936 Physics , 7 , 375 .
A new Coriolis perturbation in the methane spectrum* I. Vibrational-rotational Hamiltonian and wave functions BY H. A. JAHN Davy-Faraday Laboratory , The Royal Institution ( Communicated by Sir Bra.gg , O.M. , P.R.S.\mdash ; Received 25 May 1938 ) INTRODUCTION It is known that the infra-red absorption band of the low-frequency fundamental vibration of the methane molecule has a rotational structure which is much more complex than one would expect for such a simple molecule ( Nielsen and Nielsen 1935 ) .
It is our purpose to show that nevertheless this complex structure can be explained on the basis of a regular * The term Coriolis perturbation is used to cover all perturbations in polyatomic molecules which arise from the interaction of the angular momentum of degenerate vibrations with the rotational angular momentum of the molecule ( cf. Teller ( 1934 ) or Johnston and Dennison ( 1935 The additional forces which arise in a rotating mechanical system were investigated by Coriolis ( Journal Polytechnique , 1832 ) ; see , for instance , Routh ( 1898 ) , Dynamics of a Particle , p. 156 or Haas ( 1928 ) , Theoretical Physics , 1 , 45 .
The gyroscopic nature of the Coriolis ( or compound centripetal ) acceleration was pointed out by Thomson and Tait ( 1879 ) , Natural 392 : see Webster ( 1912 ) , Dynamics , p. 320 .
H. A. Jahn tetrahedral model for the molecule .
We shall see that the rotational levels of the vibration are perturbed by the rotational levels of the next nearest vibration in such a way as to produce in the spectrum just the observed complex structure .
The perturbation arises from a Coriolis ( or vibrational gyroscopic ) interaction between the rotational-vibrational levels of the two different modes of vibration .
In this first part we derive these Coriolis coupling terms in the vibrational-rotational Hamiltonian and find also the correct rotational-vibrational wave functions with which to carry out the perturbation calculation .
In Part II we evaluate the matrix elements of the perturbation and determine the theoretical rotational energy spectrum of .
In the third and final part we use this energy spectrum to calculate the optical spectrum and compare this with the observed spectrum .
For this purpose we calculate the theoretical intensities of the rotational fine structure lines , taking into account the nuclear spin weights of the four equivalent hydrogen atoms .
From this theoretical spectrum we calculate the theoretical envelope which would be observed with slit widths of approximately cm .
as used by Nielsen and Nielsen .
This theoretical envelope is found to agree remarkably well with the experimental envelope , even without taking into account any vibrational or rotational change in the equilibrium configuration .
1 .
PURE DEFORMATION AND ORTHOGONAL VALENCY MODES Before we can make any explicit calculation of the rotational-vibrational levels of the methane molecule we must first of all know the fundamental modes of vibration .
In the following we make use of what we call the pure deformation and orthogonal valency modes ( cf. Mecke 1930 ) .
These are not exact normal modes of vibration of the molecule , but we shall find that they are very good approximations to the true modes .
They are based on the experimental fact that the energy required to stretch a C\mdash ; H bond is considerably greater than the energy required to change the angles .
Thus we can obtain a good approximation to the low-frequency modes by introducing the condition that all the C\mdash ; H distances remain invariant .
This removes four degrees of freedom and enables us to find five different orthogonal deformation vibrations in which only the angles change .
The remaining four high-frequency modes are then determined simply by the condition that they should be orthogonal to these deformation vibrations .
These high frequency valency vibrations will involve essentially changes in the C\mdash ; H distances ; they will also , however , involve to a slight extent A new Coriolis perturbation in the methane spectrum 471 changes in the angles .
In finding these modes it will help us considerably to make use of the fact that three normal modes of vibration of the methane molecule , for any force system whatever ( consistent with the tetrahedral symmetry ) , are determined completely by symmetry .
These are the totally symmetrical vibration of type ( in the notation of Tisza 1933 ) and the twofold degenerate set of vibrations of type E. That these are completely determined is a consequence of the group-theoretical result that the methane molecule possesses one and only one of each of these irreducible types of vibration .
We denote the vibration of type by and two suitably chosen normal modes oftype by and .
In each of these modes ofvibration the carbon atom remains at rest , in the hydrogen atoms move radially in phase either away or towards the carbon atom ( so that this is a pure valency mode of vibration ) , whilst in and the hydrogen atoms move on a sphere about the carbon atom , so that the C\mdash ; H distances remain unaltered .
These latter give us therefore at once two of the deformation modes of vibration .
In the three remaining deformation modes the carbon atom takes part in the vibration .
These modes can be arrived at by regarding them as compounded out of an infinitesimal translation of the whole molecule , followed by a backward displacement of the hydrogen atoms tangential to the equilibrium sphere drawn with its centre at the displaced position of the carbon atom .
This will clearly satisfy the condition that the C\mdash ; H tances remain unchanged , since neither of the two compounded displacements change them .
The directions of displacement of the hydrogen atoms tangential to the sphere are determined completely by the conditions of orthogonality to the translations , the rotations and the twofold degenerate vibrations .
In the three remaining valency modes the carbon atom also moves and the hydrogen atoms move out or in radially towards the equilibrium position of the central atom ( and not towards the carbon atom itself ) so that the valency angles do change slightly .
In the above way we found the following set of valency and deformation modes of vibration of the methane molecule .
We use here the notation used previously ( Jahn 1935 ) : each mode of vibration is described by a set of vectors , , giving the displacements from equilibrium of the five atoms .
( The index refers to the carbon atom , and the four hydrogen atoms are numbered as in the reference just mentioned .
) These displacement vectors are expressed as linear combinations of the vectors drawn parallel to the twofold axes of the equilibrium tetrahedron .
H. A. Jahn ( a ) Deformation modes ( CH distances do not change ) .
, ; , , .
( b ) Valency modes ( orthogonal to deformation modes ) .
; , , .
A new Coriolis perturbation in the methane spectrum Here mass of hydrogen atom , mass of carbon atom , total mass of molecule .
The normalization factors have been so chosen that the reduced masses of the vibrations are all equal to , the mass of the hydrogen atom : Hence if we introduce coordinates to describe a general displaced configuration by where is the equilibrium configuration , then the vibrational kinetic energy has the form From group theory it follows , since the sets of displacements and are both transformed in the same way according to the representation of the tetrahedral group by the Wigner symmetry operators , that the potential energy has the following form : .
If the above set of displacements were true normal modes , would be zero .
This is not the case , but it can be shown that it is small compared with the other constants .
* That these displacements do form a good approximation to the true modes can be shown further by calculating the maximum angular momenta ( -values ) of the two threefold degenerate modes of vibration .
It has been * [ Note odded in proof .
] The positive value of Rosenthal 's general force constant ( Phys. Rev. 45 , 538 ( 1934 ) ) corresponds to a small value of the cross term , whilst the negative value corresponds to the large value Dennison and Johnston ( Phys. Rev. 47 , 93 ( 1935 ) ) have shown that the positive value of is required to explain the spectrum of , thus giving conclusive evidence that the valency and deformation modes are good approximations .
For the true modes , using the small value of , we find It should be pointed out that a positive value of Rosenthal 's constant corresponds to a negative value of Johnson and ( Phys. Rev. 48 , 868 ( 1935 ) ) }tant e Vol. CLXVIII .
A. H. A. Jahn shown ( Jahn 1935 ; cf. also Eckart 1935 ) that the three components of the angular momentum of the vibrations are given by , where the coefficients or are the scalar products being the set of displacement vectors obtained from by replacing each vector of by the vector product of the unit vector with and similarly for and .
Comparison with Teller 's ( 1934 ) definition of the -values of the threefold degenerate modes of vibration shows that The simple calculation gives at once ' in agreement with Teller 's result that the sum of the -values must equal independent of the force system : Substituting the values for the masses , we find The experimental values are ( Childs 1936 ) This shows that the modes we use do approximate to the true modes of vibration of the molecule .
( It should be noted that the formulae given above allow us to calculate the -values for other tetrahedral molecules for which valency and deformation modes are also good approximations .
Under this assumption we find for and for A new Coriolis perturbation in the methane spectrum With the above explicit expressions for the modes of vibration we can calculate , using a method already developed ( Jahn 1935 ) , all the Coriolis coupling terms occurring in the rotational-vibrational Hamiltonian of the molecule .
We want in particular to calculate the Coriolis coupling between the fundamentals and The Coriolis coupling term in question can be written in the form where is the moment of inertia , are the components of the total angular momentum of the molecule and are components of vibrational angular momentum obtained by combining the displacements of with the momenta of and vice versa .
( Cf .
Teller 1934 , who considers , however , only the coupling of vibrations of the same symmetry type .
) These components of angular momentum are included in the general expression given above for the angular momentum of the vibrations .
The only coefficients which do not vanish are the following We thus find the following expressions for the components of vibrational angular momentum occurring in : , with H. A. Jahn The Coriolis coupling term is an operator which acting on the vibrational-rotational wave functions of converts them into vibrationalrotational wave functions of and vice versa .
It can thus produce a perturbation in the rotational levels of and .
Now Teller 's Coriolis interaction term of the angular momentum of the vibration alone with the total angular momentum of the molecule has spherical symmetry and consequently does not completely remove the degeneracy of the rotational levels of .
As is well known it causes each rotational level ( except the level ) to split into three Coriolis levels ( -levels ) .
The term however , has only tetrahedral symmetry and consequently can partly remove the degeneracy of these Coriolis levels .
This causes a splitting as well as a displacement of the rotational levels and gives rise to the complex rotational structure of Since the term is totally symmetrical with respect to the tetrahedral group ( as can easily be verified ) , it can only have non-vanishing matrix elements between wave functions which transform according to the same irreducible representations of .
It is our purpose in the following to find those correct linear combinations of the vibrational-rotational wave functions of which transform according to such irreducible repre- sentations .
We shall carry out this explicit calculation up to the tenth rotational quantum number .
3 .
WAVE FUNCT10NS F0R TELLER'S CORIOLIS LEVELS We denote the vibrational wave functions of which correspond to one quantum of vibration in the configurations by .
It will be convenient to introduce the linear combinations These functions transform according to the odd representation of the group of all rotations and reflexions in the space of the threefold degenerate vibrations .
The rotational wave functions of the spherical top transform , for even and odd values of respectively , according to the representations of the group o - all rotations and reflexions about axes fixed the molecule ( cf. Jahn 1935 ) .
Consequently the vibrational- A new Coriolis perturbation in the methane spectrum 477 rotational wave functions transform according to the product representations ( even ) , ( odd ) .
The reduction of these product representations gives us the wave functions of the three Coriolis levels into which each rotational level is split by the Coriolis interaction term .
We denote the corresponding levels by respectively .
The formulae giving the explicit reduction of these product representations are well known and can be written as follows : , , .
( We have omitted the magnetic quantum number , since it does not enter into the transformation formulae ; in particular we could take for which case the rotational wave functions become simply spherical harmonics .
None of the Coriolis interaction terms remove the -fold degeneracy which is described by the quantum number ) These vibra- tional-rotational wave functions diagonalize the interaction term One finds easily for the matrix elements ( cf. Johnston and Dennison 1935 ) : H. A. Jahn Since these Coriolis levels have spherical symmetry ( i.e. transform according to irreducible representations of the full group of all rotations and reflexions ) , the number and degeneracy of the sublevels into which they can be split by the tetrahedrally symmetrical perturbation are determined by the reduction of the representations into irreducible representations of the tetrahedral group .
The general result of this reduction is easy to calculate and is given below , together with an explicit reduction carried out up to 4 .
TETRAHEDRALLY IRREDUCIBLE HARMONIC FUNCTIONS The reduction of the representations into irreducible representations of the tetrahedral group is determined by the characters of the former representations for the group elements of .
These characters are as follows : From these characters and from the characters of the irreducible representations of one finds the reductions given in Table I. General formulae for any value of can easily be found but are not necessary for our purpose .
Our results include those of Cundy ( 1938 ) who has given the reduction of for even and of for odd .
To carry outthe explicit reduction we makeuse of the sphericalharmonics , which transform according to for even and according to for odd .
Having reduced the representations subtended by these functions , it is straightforward to obtain the reduction in the remaining cases ( i.e. for odd and for even ) .
We follow Maxwel ] ( 1892 ) ( cf. Elert 1928 ; Bethe 1929 ) and express the spherical harmonics as multipoles having their axes along the and axes .
Their transformation properties with respect to the tetrahedraLgroup are then easy to investigate .
We replace the usual complex spherical harmonics ( as defined for A new Coriolis perturbation in the methane spectrum 479 instance by Bethe 1933 ) for each value of by the following real and orthogonal functions : .
discussion ( loc. cit we find the following expressions for these functions in terms of multipoles having their axes along the and directions : , H. A. Jahn .
where is the normalized multipole directed along the axis defined by and are derived from functions which are linear combinations of multipoles having their axes directed along the and directions .
These latter quantities are determined by the following expansions given by Maxwell : The functions are derived from and respectively by replacing in the above expansions each factor by the multipole The functions and are all normalized to From these formulae we have derived expressions for the real spherical harmonics in terms of the multipoles with up to .
The results are shown in Table II .
Since the multipoles transform like the products , it is easy to see how these functions transform under the symmetry operations of the tetrahedral group .
We denote the linear combinations of the real spherical harmonics which transform according to the irreducible representations and of respectively as follows : If two or more representations of the same type occur in the reduction of , we add superscripts to distinguish the functions , e.g. , .
From the above expressions for the functions in terms of multipoles we found that the combinations given in Table III transform according to irreducible representations as shown .
In the case of the threefold degenerate A new Coriolis perturbation in the methane spectrum 481 representations and we list only the functions , respectively , since these will be sufficient for any perturbation calculation .
As is shown in Table III the functions , belonging to the representations and are given directly without forming any linear combinations .
The general result is that for even the functions are equal to , , respectively , and for odd the functions are equal respectively to 5 .
WAVE FUNCTIONS FOR THE CORIOLIS SUBLEVELS The above reduction enables us at once to form the proper combinations of the vibrational-rotational wave functions of the three Coriolis levels which transform according to irreducible representations of the tetra- hedral group .
For this purpose we define real wave functions in the same way as we did for the spherical harmonics by , with or For the levels the proper linear combinations of these functions are just the same as for the corresponding spherical harmonics ; for the levels , however , which transform according to when is even and according to when is odd , those linear combinations which transform according to for the spherical harmonics give us the correct linear combinations of the vibrational-rotational wave functions for and vice versa .
Similarly the representations and are reversed and also the and -functions of the representation .
In this way we found the linear binations shown in Table .
These functions are the correct linear combinations to use in any calculation of a tetrahedrally symmetrical perturbation ( except when two or more representations of the same type occur ) .
H. A. Jahn 6 .
CONCLUSION The vibrational-rotational wave functions given in Table will be used in the second part of this paper to evaluate the energies of the corresponding levels when the Coriolis coupling term is taken into account .
We shall derive general formulae for the matrix elements of between the real vibrational-rotational functions , , of and the corresponding real vibrational-rotational wave functions of .
These will then be used to calculate the matrix elements between the proper linear combinations .
When two or more representations of the same type occur , the perturbed levels are found by diagonalization of the corresponding perturbation matrix .
It should be pointed out that the functions given in Table can be used also in evaluating other perturbations , for example second order Coriolis effects involving anharmonicity , or in calculations of the rotation and vibration of methane in the solid state .
The author is indebted to Sir William Brag , O.M. , P.R.S. , and the Managers of the Royal Institution , for the faciIities afforded at the DavyFaraday Laboratory where this research was carried out , and to Dr W. H. J. Childs , without whose valuable advice and discussion the work could not have been undertaken .
SUMMARY The theoretically possible sublevels into which the rotational levels of a threefold degenerate mode of vibration of methane can be split by a tetrahedrally symmetrical perturbation are given up to the tenth rotational quantum number , together with the correct vibrational-rotational wave functions with which to carry out the perturbation calculation .
Good approximations to the true modes of vibration of the methane molecule are given and used to evaluate certain Coriolis terms in the Hamiltonian which couple the twofold and threefold degenerate modes of vibration .
Table III which gives the tetrahedrally irreducible harmonic functions and extends calculations of Bethe ( 1929 ) from the fourth to the tenth quantum number will be useful in a number of connexions .
REFERENCES Bethe , H. A. 1929 Ann. Phys. , Lpz .
, 3 , 133 .
\mdash ; 1933 Handb .
Phys. 24 , 1 , 551 .
Childs , W. H. J. 1936 Proc. Roy .
Soc. A , 153 , 555 .
A new Coriolis perturbation in the methane spectrum 495 Cundy , H. M. 1938 Proc. .
Soc. , 164 , 420 .
Eckart , C. 1935 Phys. Rev. 47 , 552 .
Elert , W. 1928 Z. Phys. 51 , S. Jahn , H. A. 1935 Ann. Phys. , Lpz .
, 23 , 529 .
Johnston and Dennison 1935 Phys. Rev. 48 , 868 .
Maxwell , J. C. 1892 " " Electricity and Magnetism 1 , chapter 1x .
Mecke , R. 1930 Z. Elektrochem .
36 , 589 .
NieIsen and Nielsen 1935 Phys. Rev. 48 , 864 .
Teller , E. 1934 Handb .
Jb .
chem .
Phys. 9 , 11 , 43 .
Tisza , L. 1933 Z. Phys. 82 , 48 .
A new Coriolis perturbation in the methane spectrum II .
Energy levels By H. A. JAHN Davy-Faraday Laboratory , The Royal Institution ( Communicated by Sir William Brag , O.M. , P.R.S.\mdash ; Received 3 June 1938 ) INTRODUCTION In the first part of this work ( Jahn 1938 ) we found the correct vibrationalrotational wave functions with which to evaluate any tetrahedrally symmetrical perturbation of the rotational levels of a threefold degenerate mode of vibration of the methane molecule .
In this second part we use these wave functions to calculate the perturbation due to the Coriolis term which couples the rotational levels of the one quantum threefold degenerate vibration and the rotational levels of the one quantum twofold degenerate vibration .
This Coriolis term was derived in Part I from the general vibrationalrotational Hamiltonian of the molecule .
To carry out the perturbation calculation we shall need those combinations of the vibrational-rotational wave functions of which transform according to irreducible representations of the tetrahedral group , and these are derived here .
We find also general formulae for the matrix elements in question and use these to calculate explicitly the perturbed rotational energy levels of up to the tenth rotational quantum number .
The rotational levels of are of course correspondingly perturbed , but the calculation of these is reserved for another publication .
A new Coriolis perturbation in the methane spectrum 495 Cundy , H. M. 1938 Proc. .
Soc. , 164 , 420 .
Eckart , C. 1935 Phys. Rev. 47 , 552 .
Elert , W. 1928 Z. Phys. 51 , S. Jahn , H. A. 1935 Ann. Phys. , Lpz .
, 23 , 529 .
Johnston and Dennison 1935 Phys. Rev. 48 , 868 .
Maxwell , J. C. 1892 " " Electricity and Magnetism 1 , chapter 1x .
Mecke , R. 1930 Z. Elektrochem .
36 , 589 .
NieIsen and Nielsen 1935 Phys. Rev. 48 , 864 .
Teller , E. 1934 Handb .
Jb .
chem .
Phys. 9 , 11 , 43 .
Tisza , L. 1933 Z. Phys. 82 , 48 .
A new Coriolis perturbation in the methane spectrum II .
Energy levels By H. A. JAHN Davy-Faraday Laboratory , The Royal Institution ( Communicated by Sir William Brag , O.M. , P.R.S.\mdash ; Received 3 June 1938 ) INTRODUCTION In the first part of this work ( Jahn 1938 ) we found the correct vibrationalrotational wave functions with which to evaluate any tetrahedrally symmetrical perturbation of the rotational levels of a threefold degenerate mode of vibration of the methane molecule .
In this second part we use these wave functions to calculate the perturbation due to the Coriolis term which couples the rotational levels of the one quantum threefold degenerate vibration and the rotational levels of the one quantum twofold degenerate vibration .
This Coriolis term was derived in Part I from the general vibrationalrotational Hamiltonian of the molecule .
To carry out the perturbation calculation we shall need those combinations of the vibrational-rotational wave functions of which transform according to irreducible representations of the tetrahedral group , and these are derived here .
We find also general formulae for the matrix elements in question and use these to calculate explicitly the perturbed rotational energy levels of up to the tenth rotational quantum number .
The rotational levels of are of course correspondingly perturbed , but the calculation of these is reserved for another publication .
H. A. Jahn 1 .
ROTATIONAL SUBLEiVELS OF The rotational wave functions of the molecule transform , for even and odd values respectively of the rotational quantum number , according to the representations of the group of all rotations and reflexions .
The one quantum vibrational wave functions of transform according to the representation of the tetrahedral group .
Thus the sublevels into which the rotational levels of can be split by a tetrahedral perturbation are given by the reduction of the product representations ( even ) and ( odd ) .
From the reduction of and into irreducible representations of the tetrahedral group ( Table I , Part I ) and from the reduction of the products of with other representations of the tetrahedral group , viz. ( Tisza 1933 ) it is easy to calculate the number and type of sublevels into which each rotational level of can be split .
These sublevels are listed in Table I. To carry out the reduction explicitly we need first of all the explicit reduction of the product representations of with the other representations A new Coriolis perturbation in the methane spectrum 497 of the tetrahedral group .
We denote quantities which transform according to the irreducible representations of the tetrahedral group as follows It is then easy to verify that the explicit reduction of the product representations in question are given by the linear combinations shown in Table II .
In the case of the more than one-dimensional representations the linear combinations have been so chosen that they transform in the same way as the original quantities , as shown in the last column .
The rotational wave functions transform under rotations and reflexions about axes fixed in the ] ecule , independently of the magnetic quantum number , like the spherical harmonics .
We define real rotational wave functions in the same way as we defined real spherical harmonics ( Part I ) by where again we have omitted the quantum number Ill. From the tetrahedrally irreducible harmonic functions given in Table III , Part I and the linear combinations given in Table II it is easy to find the proper linear combinations of the vibrational-rotational wave functions of which transform according to irreducible representations .
These are given ) to in Table III , where the two independent vibrational wave functions of are denoted by and .
As in Table , Part I we list in the cases of the representations and only one of the degenerate functions , the types , and , since these will be sufficient for any perturbation calculation .
The wave functions here listed will be useful in calculations of the perturbations of the rotational levels of any vibration of type ( i.e. not merely of those of the fundamental but also of those of any overtone subvibrational level of type ) .
A new Coriolis perturbation in the methane spectrum 499 2 .
GENERAL FORMULAE FOR THE MATRIX ELEMENTS We found in Part I the following expression for the Coriolis coupling term : , where It is convenient to replace the normal co-ordinates by dimensionless quantities defined by Putting , one finds easily that Now the normalized wave functions of a harmonic oscillator are given by with ' where is the nth Hermitean polynome .
Here again , where H. A. Jahn is the frequency , the reduced mass and the normal co-ordinate .
For these wave functions we have Let us denote the zero-point wave function of the vibration by and that of the vibration by .
The two independent one quantum vibrational wave functions of we denote by and , and the three independent one quantum vibrational wave functions of by and .
From the formulae just given we find then From these one calculates easily Thus putting we find , omitting the zero-point wave functions , A new Coriolis perturbation in the methane spectrum 501 Ifwe put we have then where The squares of the matrix elements of are then given in terms of the squares of the matrix elements of by Putting in the numerical factors cm .
, we find cm .
We proceed now to deduce general formulae for the matrix elements of the operator .
Putting we have Putting as before ( Part I ) we find easily the other products not given being zero .
Further we have ( cf. Jahn 1935 ) From the general formulae given in Part I for the vibrational-rotational wave functions , and of the three Coriolis levels H. A. Jahn of we find the following general formulae for the matrix elements of the operator .
In the case of the functions a common phase factor has been omitted .
General formulae for the matrix elements branch levels .
A A branch ( levels ) .
A $/ A new Coriolis perturbation in the methane spectrum 503 branch ( levels ) .
A where 3 .
MATRIX ELEMENTS FOR THE SUBLEVELS From the general form given above it is possible to calculate the matrix elements of the operator between the wave functions for the Coriolis sublevels of listed in Table , Part I and the wave functions for the corresponding rotational sublevels of given in Table III below .
These matrix elements have been evaluated up to and are listed in Table below .
Those matrix elements which are not listed vanish .
As an example as to how the matrix elements listed in Table were derived and to explain the notation , we take the level .
We find from Table , Part I ( 7 ) .
Hence from the general formulae given above we find H. A. Jahn This expression must be a linear combination of the functions ( 7 ) , ( 7 ) which according to Table III are defined by ( 7 ) ( 7 ) Putting we find the following equations for and : These equations are consistent and give the values ' i.e. This equation is expressed in the Table by putting In every case the consistency of the equations was verified as a check on the calculations .
A further check will be mentioned below .
When two or more sublevels of the same symmetry type occur in a rotational level of we use additional suffixes .
Thus in the case of the levels we put the matrix elements being zero .
A new Coriolis perturbation in the methane spectrum 505 4 .
ENERGIES OF THE SUBLEVELS OF To calculate the energies of the individual sublevels of we apply second order perturbation theory .
If there is no accidental degeneracy , i.e. if no two sublevels of the same symmetry type occur , the result is well known : denoting by the unperturbed energy of the rotational level of , where takes into account the splitting caused by Teller 's Coriolis term and by the unperturbed energy of the rotational level of with which the level of is coupled , then the perturbed energies of the two levels are given by the roots of the determinant where is the coupling matrix element .
Putting we find which gives , neglecting compared with If there is a twofold accidental degeneracy in both the perturbing levels , then the two perturbed levels of are given by the two lowest roots of the determinantal equation Putting again we find where Neglecting compared with , we find the quadratic equation Vol. CLXVIIL : H. A. Jahn which gives us the perturbations of the two levels of .
Now it is easy to verify that these approximate roots are also the two roots of the following determinant i.e. the eigenvalues of the Hermitean square of the matrix of the coupling coefficients This is a general result holding for any degree of degeneracy ( cf. E. Wigner 1931 , p. 50 ) : the approximations to the perturbed levels are found by diagonalization of the Hermitean square of the matrix of the coupling coefficients .
In the general case and if , , are the eigenvalues of this square of the coupling matrix , then the approximations to the perturbations are where .
We have used these approximate values since they are sufficient for our purpose , but they could of course be improved upon by direct diagonalization of the original energy matrix .
Since in the levels up to never more than three representations of the same type occur in the rotational sublevels of , we have not had occasion to solve equations of higher order than the third .
In calculating the perturbation we must take into account the fact that owing to Teller 's Coriolis coupling terJn H the energy difference between the pairs of levels is constant only for the branch levels and varies A new Coriolis perturbation in the methane spectrum 515 with for the and branch levels .
Thus the unperturbed levels of are given by , and the unperturbed levels of by ( branch ) ( branch ) , ( branch ) We neglect any vibrational change in the moment of inertia , i.e. we put .
Then if denotes the frequency difference , we must multiply the calculated perturbations by the factor for the levels and by the factor for the levels .
We see that this correcting factor reduces the perturbation in the branch and increases it in the branch .
Since we are using only approximations to the true modes ofvibration and further since the frequency is not known exactly , we have taken a rounded off value of - for the factor with which to multiply the squares of the matrix elements to obtain the uncorrected perturbation cm .
In calculating the corrections to the and branch levels we have taken The results of the perturbation calculation carried out in this manner are shown in Table , both the corrected and umcorrected perturbations being given , in cm .
We see that the correction is important .
The sublevels of the individual rotational levels of as here calculated are plotted on a common wave number scale in fig. 1 .
We notice fron1 this diagram that the pattern of perturbed levels is similar ( although different in magnitude ) for the and branch levels , whilst for the branch levels the pattern is just reversed .
This fact gives us a useful independcnt check on the We also see that the perturbation is greatest for the branch levels .
where for some of the sublevels are perturbed by as Dluch cm .
H. A. Jahn CONCLUSION We see from this calculation that the Coriolis coupling term has a very appreciable effect on the energy spectrum and in Part III , where we calculate the optical spectrum , we shall see that this splitting and displacement of the rotational levels is just what is required to explain the observed complex rotational structure of the infra-red absorption band of the fundamental vibration SUMMARY Using the vibrational-rotational wave functions derived in Part I , the perturbations of the rotational levels of the fundamental vibration of the methane molecule , caused by the Coriolis term in the Hamiltonian which couples the rotational levels of with the rotational levels of , are calculated explicitly up to the tenth rotational quantum number .
Table III , which gives the wave functions for the tetrahedral rotational sublevels of , will be useful in calculating the perturbations of the rotational levels of any fundamental or overtone subvibrational level of symmetry type in a tetrahedrally symmetrical molecule .
REFERENCES Jalm , H. A. 1935 Ann. Phys. 23 , 529 .
\mdash ; 1938 Proc. Roy .
Soc. A , 168 , 469 .
Tisza , L. 1933 Z. Phys. 82 , 48 .
Wigner , E. 1931 " " Gruppentheorie und Quantenmechanik Braunschweig .
The nistry of Globulin .
By WILLIAM .
Melbourne .
( Communicated by Dr. C. J. Martin , F.R.S. Received July 26 , \mdash ; Read December 6 , 1906 .
) The object of the present paper is first to establish simple formulae for the more important of the experimental results obtained by Hardy* and Mellanby , then to interpret these in their bearing upon the chemistry of globulin in connection with a theory of colloids , and finally to find the molecular mass of globulin .
The work will be taken according to the following table of contents .
1 .
Formula for the solution of lobulin in solutions of neutral salts , and the laws associated with this formula .
2 .
Solution of globulin in acids and alkalies . .
Formulae for the precipitation of globulin by salts and by acids .
4 .
A theory of the colloidal state .
5 .
The electric conductivity of globulin solutions .
6 .
The molecular mass of globulin .
1 .
Formula for the of Globulin of Neutral Salts , and the Associated with this Formula .
Mellanby sets forth his results in graphs , from which he makes some broad deductions , to be discussed later in this section .
His most striking result , .
obtained also by Hardy , is that the fraction ( or percentage ) of a globulin suspension which is dissolved a salt solution of giyen strength is nearly independent of the strength of the suspension .
other words , the amount of globulin dissolved by a given salt solution is proportional to the amount of rlobulin presented for solution in the form of a snspension .
This fact is of fundamental importance for the colloidal globulin of the suspension .
For example , a -per-cent .
solution dissolves from a -per-cent .
suspension of obulin in 100 grammes , from per cent. from per cent. , and from per Let be the strength of a suspension when it is .
instantaneously distributed through a solution of a sadt of concentration and being iven in rammes per cubic ceIltimetre .
Let be the grammes of lobulin per cubic centimetre dissolved from the suspension when saturation is reached , and let be concentration of the salt 'Journal of Phyhiology , ' , vol. 33 , p. 251 .
, .
338 .
The Chemistry of Globulin .
equired to just dissolve the whole of G. Denote by , and by then Mellanby 's results are expressed by the equation , ( 1 ) in which A is a parameter having characteristic values for each type of neutral salt .
To illustrate the applicability of this equation , the following comparison is made of the values of for the different of and given by Mellanby for his first set of data for .
C. was not actually measured by Mellanby , but is obtainable from his first graph with only a small extrapolation as .
Hence , from Mellanby 's tabulated values of , we obtain the values of , given in the table along with his values of .
The last row contains the ratio .
The first two values are aberrant , because and being small are powerfully affected by experimental uncertainty .
If we take , and use the equation to calculate from , we find for in these first two cases and in place of and the differences being within the limits of experimental error .
Mellanby shows that the graphs for , and coincide when expressed as gramme equivalents is made abscissa and ordinate .
Hence the value of A just found for , namely holds for KC1 and Mellanby generalises this result for all neutral salts composed of two nonovalent ions , but in the appendix to his paper he points out that the graphs for , and KI do not coincide with that for , a resuIt previously found by T. B. Osborne , and given also by Hardy .
Mellanby finds that a single graph will suffice for , and .
For this type , namely , a divalent and two monovalent ions , I find , say , and for the type One of Mellanby 's discoveries leads towards a useful eneralisation in connection with A. He measures the average solvent power of any salt for globulin up to the point when the whole of a suspension is dissolved , and finds this proportional to the sum of the squares of the valellcies of its ions .
For example , if the solvent power of Na is and of Cl is , that of will be 2 , that of will be or 6 , and that of or 8 .
It seems simpler to measure solvent power by the reciprocal of the conMr .
W. Sutherland .
[ July 26 , centration required just to dissolve the whole of a given globulin suspensionFrom Mellanby 's graphs we find this concentration to be for salts of the type being the gramme equivalent .
For the type and it is and for E. Hence , for for these three types , we get the values , and If we use ramme molecular mass instead of gramme equivalent , these become , and , in which the numbers are as .
2 : 6 : 7 , corresponding to the ratios 2 : 6 : 8 found by Mellanby when using solvent powers .
We can regard this result in a different way .
Let .
us write ( 1 ) in the form and study for each type , thus making the absolute concentration prominent .
The values of for the three types are and , the numerical parts of which are , and are more nearly as 2 : 6 : 8 .
From this point of view , we find Mellanby 's law apply better to than to , to an average value of which Mellanby applied it .
In Section 4 , I shall try to give a theoretical explanation of Mellanby 's law applied to .
Meanwhile , to complete the discussion of experiments bearing on , we shall consider his results concerning the influence of temperature on the solvent power of for globulin .
He comes to the conclusion that the solvent power is proportional to the electric conductivity or summed ionic velocities of the solution of at the temperatures .
used .
But it will be better to investigate the effect of temperature on A and , and then follow Mellanby in taking account of ionic velocities in this connection .
The following contains the values of A which I have calculated from Mellanby 's data*at the five temperatures , also the ] of derived from his graphs .
The values of are also to give relative values of .
The column marked gives the viscosity of water at the five temperatures , and the last column 1906 .
] The Chemistry of Globulin .
The value of A attains a minimum between and , probably between and .
It is worth noticing that this minimum comes near to the temperature of warm-blooded animals .
The viscosity of water controls the velocities of an ion at different temperatures and is inversely proportional to them .
So the column headed gives relative values of the ratio of to the velocities of Na and Cl at the five temperatures .
These show a minimum about and can be expressed by .
( 3 ) This equation makes the value of at different temperatures depend not only on the viscosity , which determines the velocities of the ions , but on specific properties of the globulin , which are a function of the temperature .
Certainly one would expect , priori , that the profound dependence of the chemical and physical properties of proteid on temperature would appear in such an equation ; one would expect the approaching oulation at to be foreshadowed .
It is again striking that the temperature of minimum is about , near that of warm-blooded animals .
These experiments have yielded in equation ( 1 ) the result that the chemical equilibrium between globulin and neutral salts is of a simple type , dependent , as far as the salt is concerned , upon ionic electric charges and velocities , but also dependent upon special properties of gobulin .
It has already been mentioned that , and KI are exceptional to a certain extent and show that other properties of salts than ionic and velocities have an influence on their chemical equilibrium with globulin .
Moreover , the solvent powers found by Hardy*for globulin from ox blood do not agree closely with those given by Mellanby for horse obulin .
For instance , Hardy finds less powerful than , whereas Mellanby finds its power 8/ 6 of that of .
Hardy finds , and as powerful as KC1 and .
Yet 's data on whole show that valency is the chief factor solvent power .
He finds and oxalates to have the same power as and an equivalent of Na citrate times as powerful as , so the gramme molecule would be times as powerful as that of .
On Mellanby 's principle it would be ) or six times as powerful .
Further investigation must clear up these points .
2 .
Globulin Acids and Alhadics .
Hardy 's experiments show that the reactions between globulin on the one hand and acids and alkalies on the other take place undel .
the usual chemical , pp. 308 and Mr. W. Sutherland .
[ July 26 , 'conditions of combination .
Whereas the amount of a globulin suspension dissolved by a certain strength of is proportional to the original concentration of the suspension , the amount dissolved by HC1 depends only on the total amount of acid which has access to the suspension .
Hence , for acids and alkalies in place of ( 1 ) , we have the relation , ( 4 ) where is a parameter characteristic of each acid and alkali .
MelIanby established the same result and shows that does not vary with temperature .
The contrast between solution in neutral salts and in acids or alkalies is complete .
Hardy proves that a given amount of globulin reacts not with equivalents , but with molecules of such acids as , and Similarly he finds that the molecule has the same solvent power as KOH , NaOH , and .
To dissolve one boramme of globulin gramme equivalent of alkali is required , and gramme equivalent of the strong acids .
These results wall be used in Sections 4 and 6 .
T. B. Osborne , from 1899 to 1901 , obtained similar results for the vegetable globulin edestin .
3 .
Formutoe for the Precipitation of Globulin and by Acids .
The first case is that in which globulin is precipitated by excess of a solvent neutral salt .
Let be the concentration of the salt and the fraction which the precipitated globulin is of the total originally in solution , then Mellanby 's graph for is given by the equation .
( 5 ) The values of given by this for the experimental values of are compared with the experimental values of in the following table:\mdash ; To interpret this equation , we start with the consideration of the value of at which precipitation just begins .
Two actions of the salt just balance here , the solvent action and the precipitant .
If , then , by increasing the concentration to , the fraction of the globulin is precipitated , an amount of solvent power proportional to is liberated , so the solvent power , which at was proportional to , at is proportional to .
This implies , p. 357 .
1906 .
] The Chemistry of Globulin .
that the whole of the additional salt is occupied in keeping the precipitate down .
This solvent power proportional to acting on the precipitate will give a rate of solution of the precipitate proportional to , and this must be equal to the rate of precipitation when equilibrium is reached .
But as is occupied in precipitate down , the simplest assumption to make is that the rate of precipitation is proportional to .
Hence , for equilibrium , , ( 6 ) where is a parameter eharacteristic of each salt .
This equation of equilibrium is of the same form as ( 5 ) .
The precipitate in this case is probably not globulin , but a compound of ] obulin and the salt .
It begins to come down at a concentration about 40 times that which produces complete solution of the original globulin .
But Mellanby 's data for cannot be expressed by an equation like ( 5 ) .
They could be represented by portions of two straight lines .
When a salt of a heavy metal is added to a solution of globulin in a neutral salt , the strength of the neutral salt being kept constant in all the comparisons , the fraction of the total globulin precipitated is proportional to the amount of the salt of heavy metal up to a certain point .
Thus , with globulin dissolved in per cent. and then precipitated with , also dissolved in per cent. , so that the total amount of lobulin in 10 .
was always the same , the fraction of the globulin precipitated is given by , ( 7 ) up to ramme per cubic centimetre , being here the grammes of per cubic centimetre of the 10 .
used .
From to .
the formula is .
( 8 ) NIellanby converts these two lines in his graph into a single curve , but with some violence to the experimental data .
For a solution of globulin containing gramme per cubic centimetre and as precipitant , up to , while from to we can write .
For the corresponding equations are up to and up to In this connection it is interesting to review the considerable alnount of analytical work done on the albuminates of the heavy metals .
* Harnack found , in 1881 , that copper albuminate precipitated from excess of albumill * See F. .
Schulz , ' Die Gross des ' Jena , 1903 .
VOL. LXXIX.\mdash ; B. Mr. W. Sutherland .
[ July 26 , contains per cent. of Cu , and by excess of Cu per cent. In 1877 , Morner prepared a copper albuminate with per cent. of Cu , and another with per cent. In 1887 , Chittenden and Whitehouse obtained copper albuminate containing from to per cent. of Cu , but found that by redissolving and reprecipitating the percentage could be raised to Now with our formula for , and the fact that Mellanby 's globulin solution contained gramme of globulin per cubic centimetre , we get , that is to say , that at the earlier of precipitation throws down 88 times its own weight of globulin .
Thus , neglecting the small amount of Cu left in solution , we can calculate that the copper globulinate of Mellanby 's experiment contained per cent. of Cu .
If all the globulin is precipitated , our equations give that of the precipitate contains per cent. of Cu , and of it contains per cent. of Cu , so that the precipitate as a whole contains per cent. These are only about one-third of Morner and Harnack 's values for the albuminates .
But the larger value , per cent. , is close to the smallest of Chittenden and Whitehouse 's , namely , .
In the same way , from our equations for Mellanby 's data , we find that the first precipitate of zinc globulinate contains per cent. of , and that if all the globulin is thrown down the compounds average per cent. of .
Now ChiGtenden and Whitehouse found per cent. of in their zinc albuminate .
So there is a correspondence between the composition of Mellanby 's globulinates and Chittenden and Whitehouse 's ates of Cu and Zn .
Thus it bears upon the present inquiry to look more closely into the albuminates of these nvestigators .
The following table contains , in the first row , the fraction of metal in unit mass of its albuminate , in the second the equivalent mass of the metal , and in the third the ratio From the mean value of the last row we find the percentage of heavy metal in its albuminate to be E. Hence the minimum molecular mass for albumin in the albuminates is or 3300 1learly .
Hence , we infer from the above rument that Mellanby 's data would with the altogether independent data for the albuminates in making the minimum 1906 .
] The Chemistry of Globulin .
possible molecular mass for globulin about 3000 .
This result will be used in Section 6 .
We shall now discuss the precipitation by acids of globulin dissolved in neutral salts .
This is a very characteristic point in the chemistry of globulin .
While a little acid dissolves globulin , it precipitates it from solution in salts .
One would expect a priori from the similarity of salts and acids that they would help one another in globulin .
Instead of that they are antagonistic .
Hardy has shown that the solvent power of two neutral salts for globulin is the sum of their separate powers .
All the more striking , therefore , is the antagonism of neutral salts and acids .
It seems to me that the explanation of this fact is to be found in this , that the acids enter into the globulin molecule in different places from the neutral salts , because the atom can play a differelIt , in organic compounds from that possible to metallic atoms .
The ordinary organic compounds are far more stable than the organo-metallic .
Moreover , the molecule of acid and of neutral salt enter the globulin not only in different places , but in places which make them incompatible with one another , unless special provision is made to keep them in position .
Each is stable in its own place .
Together they tend to instability .
Hence , acids and neutral salts inhibit one another 's action in dissolving globulin .
We can apply this idea quantitatively to Mellanby 's experiments in which acid is made first to precipitate lobulin from solution in salts , and then to redissolve the precipitate .
He used.his solutions in such a way that the total amount of globulin and of neutral salt in the 10 .
of mixture under observation remained constant in a set of experiments .
The strength of the neutral salts was , or gramme per cubic centimetre , and of the globulin sufficient to saturate the salts .
The amount of acid present in any case in the 10 .
is specified by the number of cubic centimetres of acid from which it could be obtained .
The results show that up to maximal precipitation the fraction of the globulin precipitated is practically proportional to the amount of acid present .
In fact , Mellanby 's graphs , in which the fraction of globulin precipitated is ordinate and amount of acid is abscissa , could be replaced by two straight lines , one running up from the origin to the point of maximum precipitation , and the other running down again to a point of zero precipitation or complete solution .
To discuss the theory of the whole action , let us denote by the cubic centimetres of acid which in the absence of salt would just dissolve all the globulin .
Let be the number producing maximal precipitation iu a given salt solution containing the lobulin .
Then would dissolve the fraction of the globulin in the absence of salt .
But the salt present is able just Mr. W. Sutherland .
[ July 26 , to dissolve all the globulin by itself , hence we may say that of the total solvent power of the salt and the acid , the salt and acid possess the fractions and respectively .
So if is the maximum fraction of the globulin precipitated by , then is the fraction held in solution by acid , and the fraction held in solution by the salt is Hence , by the presence of the acid the salt is deprived of of its solvent power , and this gives us a measure of the inhibition .
The amount of acid required to inhibit all the salt will be .
But to dissolve all the globulin as well as to inhibit the salt we shall need an additional of acid .
Let be the total experimental amount of acid that will just give a clear solution again , then .
( 9 ) This is an equation for calculating from the observed values of and .
For a solution containing gramme of per cubic centimetre , with HC1 and as acids , Mellanby found and , while .
Solving ( 9 ) for the positive root of , we get .
From all his data the following table has been prepared for four salt solutions of the ength specified:\mdash ; These strengtbs of solutions dissolve nearly equal amounts of globulin in 10 , but not quite equal .
If we calculate the relative amounts of globulin by the principles of Section 1 , we find that they are as and .
Hence , relative values of , referred to equal amounts of globulin , will be derived from the values of in the table by dividing by these numbers , giving 17 , , and 9 .
Now these are to one another nearly as the values of in the table , a fact which confirms in a general way the above reasoning and calculation , though , strictly , instead of comparing 1906 .
] The Chemistry of Globulin .
with , we should compare wit referred to equal mass of globulin for each salt .
In a broad way tends to be proportional to It is a striking fact that though the acids HC1 and act by equivalents in each case , the same number of equivalents is not required in different cases ; in other words , for different solvents and a constant amount of globulin is not a constant , although with pure globulin these acids react in definite proportions .
It seems , indeed , as the precipitate we are now cannot be pure globulin .
So it is noteworthy that the amounts of acid required to yield are nearly proportional to the cube of the total globulin strength in each of the above four cases , for , taking the values of from Table and dividing them by the cubes of the relative globulin strengths given above , we get , and , or , and .
It would appear , then , that the equation of chemical equilibrium between acid and total globulin dissolved in neutral salt , is of the form .
( 10 ) Here , evidently , represents the mass action amongst three globulin compounds , whose amount in each case at the point of maximal precipitation is proportional to the total amount of globulin present .
Of these three we know that two are probably the compound of globulin with acid alone and the compound with neutral salt alone .
The third , then , is probably a compound containing both acid and salt , stable only when the other two are present in excess .
These compounds form by their joint action the precipitate , whose rate of formation is measured by .
Then in ( 10 ) must represent the rate of action of the acid tending to dissolve the precipitate , which acts in the usual way with a constant active mass .
It is worth noting that when and are the solvents , acids precipitate globulin nearly equally , while when and are the solvents , acids precipitate nearly equal fractions , but smaller than in the previous case .
It seems as if ions of the same valency , like Na and Cl or and , can co-operate better than dyad with monad as found in and .
This recalls the observation of Picton and Linder , that ions of different valency inhibit one another in precipitating colloids , while ions of the same valency co-operate .
4 .
A of State .
The characteristic of the colloid state is that the molecules cease to have a separate existence ; they link on to one another by means of the atomic electric , thus forming the meshes so characteristic of colloids Mr. W. Sutherland .
[ July 26 , Each particle in a suspension of a globulin mjght , therefore , be called a molecule , but with no advantage .
In each such particle , however , a certain pattern is repeated in three-dimensional space .
In a colloid this pattern is the real subject of the chemist 's study , just as the molecule is in crystalloids .
Because of its importance , I propose to call this pattern a semplar .
An imaginary case will enable the theory to be presented in a definite form .
The nitrogen of acts trivalently , but it can act also with pentad valency .
In the old phraseology two of the bonds of in were said to satisfy one another .
Now , in Hehllholtz 's theory , valency is the number of electrons associated with an atom .
I have proposed to consider variable valency as due to the existence of opposite electrons in the atom showing i for convenience of notation and printing I have proposed to denote the positive electron by and the negative by Thus both the triad and pentad valencies of are expressed by writing it , because and can coalesce to form a doublet , in which each electron almost completely neutralises the chemical activity of the other .
On the same principle the symbol of the hydrogen ion is , so that in there are three doublets chemically effective and one ineffective .
I have shown that these doublets and others are the cause of cohesion and rigidity .
In a crystal of frozen the molecules are all separate , each affecting its neighbours by the action of these doublets upon one another .
But imagine each ineffective doublet of to be made effective , so that reaches out to unit , with of a neighbour , then every molecule of becomes chemically linked with two of its neighbours , it becomes a semplar , and the whole crystal becomes a mesh of such semplars .
Under certain conditions it could retain its crystalline form .
This imaginary example explains the nature of a colloid on the present theory , and shows how the transition from crystalloid to colloid can be continuous with occurrence of crystalline colloids .
The formation of polymers is due to ineffective doublets becoming effective , though enerally the linking up of molecules is restricted to , roups of molecules , where measures the amount of polymerisation .
The most important case is that of water , which I have shown to be a mixture of trihydrol , found pure in ice , and of dihydrol .
In water and other polymers we have the simplest instances of semplars .
When , the measure of polymerisation , becomes large and indefinite , it gives us the colloid state , in which chemical forces between bour semplars produce effects of the same order as those between the separate doublets of molecules in the ; olid state .
Before : PhiL Mag [ 6 ] , vol. 3 , p. 161 .
Phil. [ 5 ] , vol. , p. 460 .
1906 .
] The Chemistry of lobulin .
applying this theory to the chemistry of globulin , it will be helpful to discuss with its aid the remarkable connection , discovered by Schulze and investigated by Picton and Linder , between the valency of ions and their coagulating power .
Hardy has shown that a similar relation holds for certain organic colloids .
A theoretical explanation of it has been offered by Whetham , * based upon considerations as to the probability of electrolytic ions of like sign coming together .
Though suggestive , this theory seems to me inadequate , because the electric repulsion between ions of like sign tends powerfully to keep them apart .
On the present theory , coagulation consists in out ineffective doublets so that they link together neighbour molecules of the coagulating substance .
Hence ionic charges must sometimes be instrumental in coagulation , because a free tending to attract in the doublet and to repel is an agent for opening out ineffective doublets .
Suppose that a free unites with from an ineffective of a molecule of coagulee , then it liberates , which will grasp the of a neighbouring ineffective , and so the ulation spreads until perhaps the other ion of the electrolyte with its electron steps in and arrests the process .
But there is another way in which coagulation can be propagated from an ion as centre .
The electric force of the ion will act on the ineffective doublets out to a considerable distance from itself , and , moreover , every ineffective doublet which is made chemically effective alters the local electric equilibrium by the change .
Each coagulated molecule is a of coagulation .
Hence we have the speed of coagulation due to an ion proportional to the electric charge of the ion and to the amount of coagulation already produced by it .
Let be the coagulum produced by electrolyte of concentration , then we have , with for valence of the effective ion and a constant , where is the amount of coagulum produced at a certain short time after the beginning of the process .
This small initial coagulation is produced mostly by the first process of direct linking , and is proportional to the total number of electrons which start the linking up , so we may write .
( 12 ) Now , in the experiments , the number of equivalents of electrolyte required to produce a certain amount of coagulation in a very short time are measured .
We may neglect and take as well as to be the same for each electrolyte .
If we call the power of the electrolyte , equation ( 12 ) makes its logarithm linear in the valency .
Schulze found ' Phil. Mag [ 5 ] , vol. 48 .
Mr. W. Sutherland .
[ July 26 , the coagulative powers of chlorides of mono- , di- , and tri-valent ions on arsenious sulphide to be as 1 : 30 : 1650 , the common logs of which are , 1477 and , which fall nearly into the linear form .
The data of Linder and Picton for the corresponding sulphates on arsenious sulphide are as 1 : 35 : 1023 , whose logs are and , which may be written .
The present theory , then , lends itself well to an explanation of this remarkable law .
In the chemistry of globulin , the prominent fact is that salt solutions dissolve from a suspension an amount nearly proportional to the original total strength of the suspension .
According to the present theory , the explanation is as follows : Just as ions by their elective force draw out the ineffective doublets of molecules , which they cause to join up as semplars , they can , under suitable energy conditions , break the doublets joining the semplars , which they liberate as molecules .
For example , if a crystal of our imaginary colloid were dissolved in , the action would be one in which the ions and Cl would break the doublets joining the semplars , and would join on to form .
This I take to be a representation of the solution of globulin in acids , the action of salts being not so simple .
It is a corollary of the coagulation of colloids by electrolytes .
The same electric actions that build up a coagulum can sometimes pull it down , especially if the operating ions can step into places which make the total chemical potential energy smaller than before .
Now the number of semplar doublets in a globulin suspension which the ions can break will either be the total thereof or a certain fraction of the total .
In the former case we should have solution of the colloid in the ordinary way , as by acids .
In the latter case the ions will only break those doublets which can be made unstable by the particular attacking force of ions used .
Hence the fraction of a globulin particlc dissolved by a given solution will have a definite value , and the amount of globulin dissolved will be proportional to the total number of particles , that is , to the concentration of the suspension .
This is the explanation of the most striking experimental result brought out by Hardy and Mellanby .
We now proceed to give the interpretation of equation ( 1 ) deduced from Mellanby 's experiments .
Let us denote the semplar of globulin by G. The simplest possible formula for the semplar at the moment of its detachment is .
Then either and can coalesce to form an ineffective doublet , as a molecule , or if ions like Na and Cl are present , their electric chal.ges will link them on , giving , or , more briefly , NaGCl .
Now the fact that about 150 times as many equivalents of as of HC1 are required to keep a iven amount of globulin in 1906 .
] The of Globulin .
solution , shows that NaGCl must be relatively very unstable ; a large excess of Na and Cl ions is required to keep it in existence .
Its existence depends on the breaking of doublets and keeping them broken .
The number of bonds broken per second by the Na and Cl ions will be proportional to the concentration of these ions and to the number of akable bonds remaining to be broken , that is to .
The rate of solution can be written .
Now the rate at which the dissolved tends to go out of solution will be proportional to and to a function of , which just vanishes when , the concentration which dissolves all the globulin .
The simplest such function is , so we write the rate of precipitation as .
So , for equilibriunl , , , which is identical with ( 1 ) and ( 2 ) .
Now A is the ratio of , the velocity of solution of the globulin , to , the velocity of precipitation .
If instead of we use the number of ramme molecules in these equations , its coefficient becomes , which boives a measure of the ratio of the velccities of solution and precipitation when gramme molecules of different salts are compared as to their actions on globulin of a given concentration .
Now the velocity of solution of ] obulin by a molecule will be proportional to the rate at which the doublets are broken by ionic charges , so it will be proportional to the velocity of the ion and its , that is to say its valence .
But in the precipitation of globulin the ions have to be extracted from combination with the lobulin , so the number of doublets to be broken will be proportional to the valence of the ions .
It is assumed here that in salt globulins an ion is loosely attached at the semplar doublet which it breaks , and that as power is proportional to valence , so also is the of attachment .
The velocity of precipitation will be therefore inversely as the valence .
Hence will be proportional to the square of the valence , which is the modified form iven to Mellanby 's law in Section 1 .
As regards the dependence of the velocity of precipitation on velocities , it will arise mostly from motion of the combined ions in the globulin compound .
We find , then , that must be proportional to the velocity of the ions and to a velocity on the globulin .
Now the velocity of the ion must not be confused with the quantity defined in physics as the ionic which is a velocity generated under a certain unit external electric force .
Apparently , in reacting with globulin , different ions move with about the same speed .
But as ards the e , ffect of change of temperature on the velocity of ions , we know the velocity to be inversely as the viscosity of water at Mr. W. Sutherland .
[ July 26 , different temperatures .
Hence , our equation ( 3 ) expresses both this dependence of on the viscosity of water and on velocity inside the globulin compound .
Thus the present theory of colloids is able to give a consistent account of these important features of the chemistry of globulin .
The precipitations of globulin hardly need review in this connection , as they have been pretty fully considered in Section 3 .
5 .
The Electric of Globulin Solutions .
Hardy , by the experimental difficulties of measuring these conductivities , has obtained data of great importance , which I propose to make more available by expressing them in interpretable formulae .
With HC11 gramme of globulin was dissolved in gramme equivalents of acid , and the solutions prepared from this by successive dilution have their strengths given by , the volume of solution containing 1 gramme equivalent of .
From the measured specific conductivity of these solutions the molecular conductivity of the HC1 is calculated in the usual way , but it should be noticed that it is expressed with reference to a unit of resistance 10 times as large as that often used , so that conductivities are made 10 times as large .
For HC1 globulin solutions the molecular conductivity of the HC1 given by Hardy*is expressed by\mdash ; , ( 14 ) as the following comparison shows:\mdash ; At infinite dilution ( 14 ) gives , the value for pure HC1 being 3840 .
It appears , then , that at infinite dilution HC1 globulin is entirely dissociated , yielding completely ionised .
The equation having so a range must have fundamental theoretical significance , which is easily brought out .
The occurrence of reminds us of Kohlrausch 's discovery that the molecular conductivity of dilute electrolytic solutions is .
a linear function of , a fact which I have sought to trace back to the 3 occurring in the formula for trihydrol .
For the molecular conductivity of an ion of radius , valency , dielectric , p. 274 .
See " " The Molecular Constitution of Water Rlld " " Ionisation , Ionic Velocities , Atomic ' Phil. Mag [ 5 ] , vol. 50 , and [ 6 ] , vol. 3 .
1906 .
] The Chemistry of Globulin .
capacity , and charge , moving through water of dielectric capacity and viscosity the formula is\mdash ; ; that is to say , the ionic velocity is inversely as , so that the time taken by an ion under standard electric force to travel 1 cm .
is proportional to Now in the case of HC1 globulin the corresponding time is proportional to , and therefore to .
Here represents the time that would be taken by the and the Cl ions to increase their distance apart by 1 cm .
, if they were always dissociated from the globulin .
The term represents the time lost on account of the presence of the globulin .
The way in which this time is lost is the following : Suppose that HC1 globulin is HGCI , yielding two sets of ions and Clb , and also GClb and H# .
So the ion travels of its path as and part as , while the Cl ion travels part of its path as Clb and part as GClb .
Let be the fraction of each second which and Clb spend in the forms HG and During this fraction each moves with the smaller velocity due to the large radius of , for the rest of the second it moves with the ionic velocity appropriate to or Clb .
The average velocity of the ion is Hence the time taken by the and Cl ions to separate by 1 cm .
is proportional to\mdash ; .
( 16 ) If , then , we write in the , we have .
( 17 ) Denote by , then we have proportional to , and therefore to , where is the concentration of the .
If is small , as Hardy directly proved it to be in the way we shall discuss in the next section , then stands almost for , the fraction of the total or Cl existing at any moment as or GClb , while stands for the amount of .
In short , represents the fraction of the HC1 which is combined , and that which is free .
Hence the proportionality proved above , which may be written , ( 18 ) is an equation of chemical equilibrium expressing that the amount of combined HC1 existing as HGCI in the form of ions HG and GC1 is proportional to the amount of free HC1 and to the amount of something else .
In " " Ionisation , it is suggested that this something in ordinary , p. 173 .
Mr. W. Sutherland .
[ July 26 , electrolysis is broken out of trihydrol .
It is shown how the number of these in a solution is proportional to , so that when they act upon the whole solute they produce an of amount proportional to that is , which , in the present case , acting on the residue of free HC1 keeps the amount of combined HC1 formed .
Hence , for equilibrium , we have as in ( 17 ) and ( 18 ) .
If in ( 16 ) we neglect and put , we find , that is to say , that , which isjust saturated with lobulin , contains 78 per cent. of the and Cl ions in combination with globulin , and only 22 per cent. free .
The experiments of Hardy , wheu summarised in ( 14 ) , a pretty complete insight into the chemistry of HC1 globulin .
As regards alkali globulins , the data of Hardy can be expressed by We shall at first discuss only the first and third of these characteristic of NaOH and .
When is infinite , the first gives , while in the same units the standard data of Kohlrausch would boive 444 for and 1740 for OH , that is .
The formula gives complete dissociation of the alkali globulin at infinite dilution .
It is noteworthy that the first term in ( 18 ) for NaOH globulin is 220 , which is half of the standard value for Na .
It appears , then , that half the unites with gobulin to form a globulinate , which dissociates completely at all ] dilutions of Hardv 's experiments , giving Na ions , and globulinic acid ions whose velocity is too small to make itself apparent .
As the OH of this half of the original NaOH does not contribute to the nlolecular conductivity in the same way as the Na , but yet appears the conductivity at infinite dilution , we see that it joins on to the lobulin iu a uilar way to that of the other half of the NaOH .
The of the lobulin displaced by the Na atom mnst also join on to the obulin in a different function .
The other half of the NaOH seems to combine with globulin in a way similar to that of .
A compound such as NaGOH to be formed ( with OH attached from the first half of the NaOH ) , yielding ions and GOH } which may be practically two val.ieties of the lobulinic acid with which the other half of the combines .
the second of ( 19 ) it appears that if twice as much NaOH is used as suffices to dissolve all the lobulin , its molecular con1906 .
] The Chemistry of lobulin .
ductivity is but little changed .
So in this case a quarter of the base reacts in quite a different way from the remaining three-quarters .
The same considerations apply to the case of .
The first term 290 is close to the 311 which the standard datum of Kohlrausch would give if half the existed as basic ions taking the place of the ions of globulinic acid .
The other half of the combines additively like the second half of the NaOH .
In the alkali globulins we must distinguish between the basic half of the alkali and the additive .
To the additive half exactly the same considerations apply as to the additive HC1 globulinate , and the degree of dissociation of the additive alkali hydrate can be determined as in the case of HC1 globulin .
Hardy found that to neutralise globulin to phenolphthalein with NaOH the amount of alkali required is twice that needed to dissolve the globulin .
Here we have additional evidence that the alkalies act in two ways .
Hardy finds that 1 gramme of globulin is dissolved by gramme equivalents of alkali , and by of acid .
Now only half the alkali displaces acidic , hence the acidic atoms are only one-quarter the number of places where HC1 can add itself to the globulin molecule .
Mellanby 's measurements on horse globulin do not agree with Hardy 's distinction between the number of equivalents of acid and alkali needed to dissolve globulin .
He finds a gramme of globulin dissolved by about gramme equivalents of HC1 and about the same number of NaOH .
Globulins of different and history appear to differ to an extent that limits us at present to a discussion only of orders of magnitude in this connection .
Concerning the molecular conductivity of neutral salt solutions saturated with globulin , Hardy made some measurements .
of concentration had its conductivity depressed per cent. , of by per cent. , and of by per cent. While of the of conductivity was only per cent. Hardy does not give the strength of the globulin suspension with which he saturated his salt solutions , but states that a ramme of globulin was dissolved by from ramme equivalents up to .
About 100 times as many equivalents of are required to keep globulin in solution as of NaOH , but as the oolobulin reduces the conductivity of about 2 per cent. , we may take 2 per cent. of the salt to be combined with globulin , so two equivalents of actually unite with globulin for one of NaOH .
Thus it appears that in its action on globulin is liker to HC1 than to NaOH .
The conductivity measurements fully confirm the conclusions arrived at from the solvent powers of neutral salts , namely , that the compounds for med are very unstable and a excess of to maintain their existence .
Mr. W. Sutherland .
[ July 26 , 6 .
The olecular Mass of Globulin .
In " " A Dynamical Theory of Diffusion for Non-Electrolytes and the Molecular Mass of Albumin , have sought to show how the molecular radius of a molecule can be found from its coefficient of diffusion , having previously obtained the relation reproduced as ( 15 ) in the present paper between molecular conductivity and radius .
These two definite and accurately measurable velocities , the ionic and that of diffusion , are convenient aids in the measurement of large molecular masses .
From the Graham-Stefan coefficient of diffusion for albumin the radius of a gramme molecule of albumin was found to be 30 cm .
, whence , from the percentage composition , the following formula is obtained for albumin , with a molecular mass 32,814 .
Now Hardy has obtained a direct measurement of the ionic velocity of globulin , namely , about cm .
per sec. , which corresponds to a molecular or rather ionic conductivity 100 in the units used by Hardy , and 10 in the units of Kohlrausch .
With the latter units ( 15 ) is transformed to , ( 20 ) in which is the dielectric capacity of the ion , having for globulin the value 2 nearly , because for proteids the index of retraction which equals is about is the volume of the molecules in a gramme molecule of the ion , and its valence .
We are going to use this equation to find for globulin , using Hardy 's value 10 for .
But so far is unknown .
It is the basicity of globulinic acid which is easily determined approximately by the facts we have in hand .
Thus it has been shown that when NaOH dissolves globulin , half of it acts as a relatively strong base .
Let be the molecular mass of globulin , then gramme molecules of NaOH act as a strong base with grammes of globulin .
But Hardy finds gramme molecule of NaOH just dissolves 1 gramme of globulin , so ramnle molecule satisfies the -basicity of globulin .
( 21 ) With equation ( 20 ) gives .
( 22 ) For the percentage composition of globulin , we have Hammarsten 's determinations , with which I shall give the limiting volume of a gramme atom of the elements taken from my ' Phil. Mag. ' papers on molecular force .
1906 .
] The hemistry of Globulin .
Let be the number of atoms of in the molecule of globulin , then , ( 23 ) and also 4 . .
( 24 ) These four equations us the four unknowns , and M. The best way solving them is to give integral values , and find which value makes the four equations as consistent as possible within permissible limits of error .
With gives ) , and then gives , which in ( 24 ) yields , and this in ( 23 ) makes As they stand , the two values of , namely 20,000 and 3362 , are inconsistent .
If , the first becomes 40,000 and the second or 26,896 .
For the co1lparison is between 60,000 and or 90,774 .
No value of above 3 is permissible .
Hence the choice for the formula of globulin is between one which makes it a dibasic acid of molecular mass about 40,000 and another which makes it a tribasic acid of molecular mass about 60,000 .
Let us inquire what in the experimental data would make the four equations strictly consistent for and .
The most uncertain number is Hardy 's valuable datum .
For a transparent opalescent dialysed solution of NaOH bolobulin he found and estimated 10 as the probable limit to which the value would rise with complete ionisation of the compound .
But his solution being opalescent , there must have been some electric endosmose such as he found with globulin , for which raised up to 23 .
In the observed for NaOH lobulin there is already some compensation of errors , and it is an open question whether the limit is more or less than 10 .
It is easy to calculate the value of in ( 20 ) which would make and satisfy .
It is .
So for and we should need .
Hence the decision between the two alternatives should be reached by a sharper detel.mination of the ionic velocity of globulin or by its coefficient of diffusion , or preferably by both of these methods .
Probably is the better value for .
With this and and we get .
Hence the formula of globulin is This formula of the globulin molecule a molecular mass times that iven above for albumin obtained from the Graham-Stefan coefficient of diffusion .
That formula can be greatly simplified for discussion if we replace the number of carbon atoms 1436 by round number 1440 , the 2364 of by 2400 , the 359 of by 360 , and of by 480 , and neglect for a while the 15 for S. These challges keep the pelcentage comMr .
W. Sutherland .
[ July 26 , position of albumin true within the limits of analytical error , and enablo us to write the main part of the egg albumin molecule as 120 .
The formula of globulin cannot be quite so much simplified , but we can write it ecting meanwhile .
The group , which is repeated 120 times in the albumin molecule , is closely related on the one hand to the polypeptides synthesised by E. Fischer , and to the pure peptones prepared by Siegfried .
For the derivation of dipeptides and tripeptides from the amino acids by abstraction of water we have the equations* COOH -NHCOCOOH , dipeptide ; COOH -NHCO)COOH , tripeptide .
If , now , in this general formula for a tripeptide we put , we obtain , a group containing 3 more than our typical group above .
If , then , we remove from this polypeptide , we obtain a trivalent radicle , , which we find repeated 120 times in the molecule of egg albumin .
It hardly matters at present how many times may be removed from one of these groups and inserted into others , for we are considering only the broad plan on which the albumin molecule is built .
Similarly , the properties of globulin depend on a certain preponderating group whose average composition is close to .
These groups are almost as closely related to the pure natural peptones prepared by Siegfried as to the synthesised polypeptides of E. Fischer .
These are the antipeptones and , or trypsin-fibrin peptones and , of which the next would be from our albumin group by OH and from our lobulin group by .
Even if these formulae need to be doubled , the relation would not be altered .
Then come the pepsin-fibrin peptones and .
It is ficant that the latter could be produced by the additive union of the antipeptones and , the experiments of W. Neumann and fried have yielded the result that in the ltipeptones t ions are to the OH ions as 2 : 1 , and in the pepsin peptones as 3 : 2 .
But the number of ions gives the number of COOH radicles in the molecule , wlJile that of the OH ions gives the number of radicles .
Hence we infer that the group when converted into an antipeptone , contains COOH twice and once .
Hence , in the up of such groups to the number of 146 to form globulin , there must be a amount of internal salt formation , since COOH occurs only two or tltree times in the complete globulin molecule .
It ' Phyp. .
siol .
' vols .
35 38 , 45 .
1906 .
] The Chemistry of Globulin .
is most probable that the internal salt formation goes on in such a way that COOH and neutralise one another .
* Thus in the extreme case of all the oxygen existing as COO in , two-thirds of the total nitrogen would be present as joined up with COOH in the globulin molecule .
When the globulin molecule is made an ion , it has probably only two COO adicles showing the acid character .
Hence we infer from our discussion of Hardy 's conductivity measurements that the globulin molecule contains eight times in such a form that HC1 can join on as it does with to for1n .
These eight and the two COOH are the ] ones left free out of about 290 of each sort existing in the lobulin molecule .
By more or less drastic treatment the latent and COOH radicles may bc called into partial action , and so and alOumin may be made to show apparently acidity and basicity than the eight and the two determined above .
ln the same way arise the discre } ) results of different experimenters , discussed in Section 3 , as to the percentage of heavy metals in their compounds with .
Different treatments open out or break up the albumin molecule differently .
But yet a rather striking reement appears in these discordant results if we regard them in the following way .
The in the percentages of Cu in albu1ninates and globulinates is in Mellanby 's globulinate , in the albuminate of Chittenden and Whitehouse , Morner , Jfarnack , Morner , Chittenden and Whitehouse , Harnack , and 11 , the of Bitthausen 's measurements with vegetable albumins .
For the weight of proteid combined with a gramme equivalent of Cu these give the numbers\mdash ; which are expressed in the second row as multiples of 291 , these multiples all nearly exact integers 1 , 4 , 5 , 6 , 8 , 9 , 15 , and 24 .
, then , we have evidence of the splitting up or out of the albumin ecule into parts , each of which contains an !
number of a group whose molecular mass is about 291 , which is quite close to the of our .
This roup with would .
This seems a fairly conclusive proof that there is a amount of intel.nal salt formation in albulnins .
The ) found in Section.3 for the mean nlass of combined with an livalent of heavy metal in Chittenden Whitehouse 's experiments sents a of onr ) .
It is to note that the treatnleltt of a etable a coppel up the lllulccule into of the size of a peptone lnolecnlc .
This prubably ' See , ' Chemistry of the , pp. 210 and VOL. LXXIX.\mdash ; B. Mr. W. Sutherland .
[ July 26 , accounts for the powerful toxicity of copper towards those lower plants to whose protoplasm it gains ready access .
The large molecular mass of about 40,000 found for globulin accounts for the fact emphasised by Hardy that the polybasic acids combine by molecules and not by equivalents with globulin .
The eight radicles are so widely distributed over the large molecule that only one can come within the reach of an acid molecule .
Thus only one OH radicle in and can operate .
In the same way , since globulin contains only two COOH radicles , these are probably far apart , and a divalent atom like Ba can replace only one H. It would be possible for Ba to unite two globulin molecules through a COO radicle in each , and for to unite two through an radicle in each .
Probably the colloidal state of a globulin suspension is due to the ordinary ineffective doublet of trivalent nitrogen becoming effective , thus the globulin molecules to grip each other by means of the eight free radicles in each .
Solution in acid occurs when the acid satisfies the extra valencies of the radicles , thus breaking their hold on one another .
The semplar of globulin escapes into the freedonl of solution as part of the molecule .
In edestin , T. B. Osborne has found what he calls the ntonohydrochloride insoluble , while the dihydrochloride is soluble .
The explanation of this is that in the monohydrochloride sufficient junctions crst the semplars are still left intact to keep the colloidal monohydrochJoride from going into solution .
The equivalent weights , 14,300 to 7000 given by Osborne for edestin , are 1/ 3 and 1/ 6 of the 40,000 found above for globulin .
As to soltion of colloidal globulin in bases , we have seen that in the case Hnrdy 's ox globulin it was not sufficient to supply lnerely to forlu the easil globulin salt , but twice as much was required for complete solution .
Here the extra NaOH is required along with OH and due to the ] lalf of the NaOH to break colloidal junctions between tho seltplars .
How it does this as efficielltly as four times the equivalent HC1 fornts sn ) sting chemical ( .
The ) robable mode of eration of in colloidal globulin has already been discnssed .
the results of and Mellan by sinlple , il is that the lutio of lobuli and its precipitation take ( conditions of .
For ) if ' is the ctio of a dissolved salt solution atio is the ction ( of is equired just to dissolve the whole 1906 .
] The Chemistry of of the suspension , we get equation ( 1 ) , in which A is the ratio of a velocity of solution to a velocity of precipitation .
Mellanby 's discovery of the dependence of on valency and ionic velocity is applied to molecular mass , and it is shown that when temperature varies , not only does depend upon the viscosity of the solvent water , but also on a function of temperature given in equation ( 3 ) which expresses the part played by globulin .
It is noteworthy that this function has a minimu1ll value about C. , near the temperature of warm-blooded animals .
For the precipitation of globulin by excess of , equation ( 5 ) is established , namely , being the fraction which the precipitated globulin is of the whole and the concentration of the solution in grammes per cubic ceutimetre .
Then follow formulae for the remarkable precipitation of globulin by acids from solution in neutral salts .
From these it appears that three compounds of globulin react in producing this precipitate .
Section 4 is devoted to a theory of the colloidal state , namely , that a colloid consists of molecules which are chemically united hbour to neighbour by the action of valencies which are usually latent .
Cases of multiple valen cy like that of and oxygen are best accounted for by the electron theory of valence due to Helmholtz , with the assumption that a single atom can contain both negative and positive electrons .
Thus the nitrogen atom has four negative electrons and one positive .
In trivalent nitrogen the positive electron neutralises one of the negative , so that the atom has only three effective negative electrons along with an ineffective doublet .
But if this doublet is opened out , it can link on its atom to two neighbour atoms , the ative electron a positive , and the positive a ative .
According to chemical theory of the colloidal state , the term molecule ceases to have a useful meaning when plied to a colloid , so the term semplar is used to name that structure which is repeated like a pattern in three dimensions through a colloid .
By suppression of the colloid-producing valencies of doublets , a mass of semplars is caused to fall into a collection of separate molecules .
illustration of the usefnlnless of this theol it is applied to show dependence of the coagulating power of an ion on its valence .
It is then applied also to explaill the remarkable fact that antount ( dissolved ) a iven salt from a suspension depends on the concentration of the suspension .
action of the of a neutral salt in globnlill is treated as another manifestation of the same effect enables them to coagulate ious snlphide .
This eory of the colloidal state leads to a theory of The Chemistry of Globulin .
equation ( 1 ) and of the laws of A in it , especially that is proportional to the sum of the squares of the valencies of the ions of the electrolyte .
In Section 5 the experiments of Hardy on the conductivities of globulin solutions are expressed by formulae which admit of very simple interpretation , and connect these conductivities with those of ordinary electrolytes .
At infinite dilution the HC1 compound of globulin is completely dissociated , its conductivity being that due to the completely ionised , the part due to the globulin being then too small to appear .
For any other dilution the amounts of combined and free HC1 can be calculated by equations ( 17 ) and ( 18 ) .
In the compounds of NaOH and with globulin the conductivities show that half of the base combines with the globulin in a way which proves it to be an acid , the salt being completely dissociated and showing the conductivity due to Na or alone , that of the globulinic acid being too small to appear .
The other half of the base which is required to dissolve globulin completely combitles with it in a way similar to that of , namely , by addition .
In Section 6 globulin is shown to have probably a molecular mass 40,000 and a basicity 2 , the alternative .
a mass 60,000 and basicity 3 .
Further experiments like those of Hardy on the ionic velocity of globulin and also on the coefficient of diffusi of globulin solutions would decide the matter , though doubtless various globulins differ in mass and basicity .
group , , related to polypeptides and peptones is shown to be the predominant structure in albumins .
The discrepant results of different experimenters on the precipitation of albumin by heavy metals fall into harmony when it is proved that they precipitated different integral nunnbers of a group such as this in combination with an equivalent of heavy

