THE COAL FIRE.-H. M. Stationery Office have published for the Department of Scientific and Industrial Research a report entitled "The Coal Fire," which describes the work carried out by Dr. Margaret Fishenden in connection with a research into Domestic Heating; the investigation was directed by the Air Pollution Advisory Board

T Chemical Manufacturers and others.—

For Immediate Sale: Six practically new 60in. water-driven Hydro Extractors, by the eminent makers, Watson, Laidlaw and Co., Ltd., Glasgow; instant delivery. Apply to JOHN H. RIDDELL, LTD., 40, St. Enoch Square, Glasgow.

of the work have been made by the Research Department. The investigation into the efficiency of open fires has yielded a collection of carefully ascertained data from which it is believed that a

Third Class Honours, desires situation as Chemical Apprentice, to train as Analytical Chemist.-State particulars to H. N. NORTH, The Cottage, Weedon, Northamptonshire.

whole question of the use of coke and other forms of smokeless solid fuel in domestic fires. The work is of so important a character that it has been arranged, in consultation with the Manchester Air Pollution Board, that the report should be published; it is hoped that in this way the work will receive the wide publicity it deserves. Copies price 4s. (by post 4s. 3d.) may be obtained through any bookseller or directly from H.M. Stationery Office, at the following addresses: Imperial House, Kingsway, London, W.C.2; 28, Abingdon Street, London, S.W.1; 37, Peter Street, Manchester.

NOTICES.

EDITORIAL.-All Literary communications and Books, Chemical Apparatus, &c., for review or notice to be addressed to the EDITOR.

SUBSCRIPTIONS, £1 12s. per annum, payable in advance, should be addressed to the MANAGER.

BACK NUMBERS and VOLUMES can be purchased on application to the MANAGER.

THE CHEMICAL NEWS,

97, SHOE LANE, LONDON, E.C.4.

ADVERTISEMENTS.

Experienced in analysis of cement and raw materials preferable. State age, training, experience and salary required to Box 813, c/o SCOTT & SON, 63, Ludgate Hill, E.C.4.

THE Proprietors of Patents Nos. 20550 of

1914 for "Improvements in the manufacture of Alcohols", 14709 of 1915 for "Improvements in Halogenating Hydrocarbons and other Chlorating Processes, and in effecting re-actions generally between Gaseous or Vapourous Bodies"; and 101708 of 1916 for "Improvements in the manufacture of Products from re-action between Corrosive and Non-corrosive Gases, and especially of Chlorine Products or other Halogen Products of Hydrocarbons, and in apparatus for use in such manufacture"; are desirous of disposing of the Patent Rights or of negotiating for the grant of Licences to working thereunder. All enquiries should be addressed to JOHNSONS & WILLCOX, 47, Lincolns Inn Fields, London, W.C.2.

dressed to

paid for any of the undermentioned numbers of the CHEMICAI NEWS which may be forwarded to this office:

THE CHEMICAL NEWS.

VOL. CXXI., No. 3155.

THE CONSTITUTION AND STRUCTURE

OF THE CHEMICAL ELEMENTS.

By HAWKSWORTH COLLINS.

THE NUMBER 23.

THE following investigation chiefly concerns the heavier elements as the simpler elements from H to Mn were considered in the CHEMICAL NEWS, December 26, 1919.

Since it has been proved: (1) that the atomic weights are integers (CHEMICAL NEWS, November 28, 1919), i.e., that all the elements are polymers of hydrogen; (2) that the atomic weights of artiads are even integers, and those of perissads odd integers, of which the only exception so far found is nitrogen (CHEMICAL NEWS, December 19, 1919); (3) that there are in general large errors in the higher atomic weights (CHEMICAL NEWS, November 28, 1919); (4) that the higher the atomic weight the greater is the probable error (CHEMICAL NEWS, November 28, 1919); (5) that when oxygen is taken as 16 the errors in general make the atomic weights too large, and when H is taken as the unit the errors make the atomic weights too small (CHEMICAL NEWS, April 9, 1920); it is evident that the experimental atomic weights referred to H as unit are as important as those calculated from 0=16 also that the integers indicated by these five proofs are not obtained arbitrarily for each one is the nearest odd or even integer to the two experimental atomic weights according as the element has odd or even maximum valency. These integers are given in Table 1.

TABLE I. Integers

Cu is known to act either as a monad or as a dyad, and it is generally supposed that the monadic valence is one of the two dyadic valences, but it will be shown later that this is not the case; so that Cu is really a triad and must be represented by an odd integer. The same remarks also apply to Hg.

Of the 83 recognised elements, the first 25 in ascending order of atomic weight have already been considered, and 12 are rare elements about which little is known and which cannot be

arranged satisfactorily in the Periodic Table. The remaining 46 are given in Table I., so that there is no arbitrary choice of elements.

In Table II., a very large generalisation is given connecting the integers in Table I., supported by four sets of coincidences.

TABLE II.

At. wt. divided into

2 portions

3......

4....

5......

6......

H-H-H-H-H-Hg 7......

Ga

Vv

Na

H-Na-H-H

69

5......

Ge

As

74.4

Se

78

63= 23+ 40

Gal

Br

Na - Svi

78=2×23+ 32

Kr

82

82.92

82.3

Rbi

Rb

Na, - KI

85=2×23+ 39

Zriv

Na - Sciv

90=2×23+ 44

Sr

88

87.63

86.97

Movi

89.33

88.63

Ruvu

Zr

Rhvi

Mo

Na-Al Na, - Mnvi Na, - Fevi

96

Cdn

107.88

114.8

113'9

117.8

Sb

Ball or Iv Ceeven Hgodd

I

Nto

Te

128

Na - Sri

La

Na - Rheven

Na2-Znu Na-Ga Na2-Gerv Na-Clvi Na, - Zriv Na - Tieven Na4 - Agodd Na - Xeo Na - Celv Na2- Ireven

The four coincidences in Table II. are:

I. Of all the elements Na is the most likely one to enter into the constitution of several elemens, as it is more widely distributed than any other.

2. Table II. connects pairs of elements, which are already known to be similar chemically, or which are especially associated with one another in mineralogy. (The latter data will be given in the next paper).

3. The number of portions (Na) added to the one of any pair in order to form the other is always such that the valency of the heavier can be explained. E.g., suppose that Zn and an element similar to it chemically had been found to have a difference of 23 in their atomic weights, it would have been impossible to explain how they could both be dyads, knowing that the electric charges are quiescent in pairs; but since Cd Na,Zn, it is evident that if the pair of valences emanating from Na, are quiescent, Cd would have the same valency as Zn.

4. The following facts, which concern a large proportion of the elements given in Table II., show that there is frequently present in the constitution of the heavier element of a pair something which is common to them all and which causes each heavier element to be more volatile than the lighter element, or which causes the former to melt or volatilise at an abnormally low temperature :—

(a) K melts at 335° abs. temperature, Rb (Na,K) at 311°, Cs (Na,K) at_300° (b) Zn melts at 692°, and Cd (Na,Zn) at 594°. (c) Ag melts at 1234°, and Hg (Na,Ag) at 234°. (d) I (NaCl) easily volatilises. (e) Na melts at 368°, and Ga (Na,) at 303°. The latter liquifies between the fingers, and cryolite (Na,AlF.) melts in a candle-flame.

(f) Ge melts at about 1173°, and Sn (Na,Ge) at 605°.

(g) In (Na,Ga) melts at 449°, an abnormally low temperature.

(h) Ce melts at a lower temperature than either Cb or Ti.

(i) Ba melts at a lower temperature than Zr. The following fact is important in this connection -All alloys except Na,K and NaK, are solid at the ordinary temperature.

The number 23 occurs 70 times in connection with the 41 elements concerned in Table II. 23 is a large number under the circumstances. It would have been a very awkward matter for demonstration if the main constituent of the elements had been represented by a small number, say, 5, for instance; but, being as large as 23 the matter becomes very apparent, and the probability of its truth is very much greater than it would otherwise be.

Although one or two of the results of Table II. may in the future be shown to be wrong, it is quite impossible that such a large generalisation as this could be evolved if there were no truth in it.

Remarks by the International Committee with regard to this year's list of atomic weights make it plain that chemists admit that large errors have not yet been eliminated. In the case of boron they say that Smith and Van Haagen find that its atomic weight is 109 instead of 11, and that they discuss all previous determinations and show

wherein they were affected by errors.

The reduc

tion of 1 to 109 is equivalent to a reduction of 10 in Cd which would bring it down from 112:4 to 1114, or to a reduction of nearly 20 in the case of Pb. Again, with regard to Zr the International Committee say that Venable and Bell by pointing out sources of error in all previous values find its atomic weight to be 9176 (not yet authorised) instead of 906.

Even if these two proposed alterations are found in the future to be incorrect, still the two admissions with regard to errors hold good.

It is not intended to be inferred that the heavier element has always been actually formed by the union of the lighter element with one or more masses of 23, but that there is a difference in their constitutions of one or more such masses. Some of the cases to which this remark specially refers are Zr, Sn, Ba, and Nt. E.g., it is not intended to be inferred that Sn has been formed by the union of Na, with Ge, because Sn is never found associated with Ge in minerals, but that the combination of Na,Ge is due to the inter-relationship of the elements.

If the nearest integer for Cb had been 94 instead of 93 it would have come into the generalisation as Na,Ti; and if the nearest odd integer for Cs had been 131 instead of 133 it would have come in as Na, Rb or Na,K. Evidence that the actual atomic weights of Cb and Cs are 94 and 131 will be given when each element 1S considered separately.

The following facts (Tables III. and IV.) could not be brought into Table II. (except those in brackets) because the pairs of elements are not obviously connected either in chemistry or mineralogy, but they are connected by the inter-relationship of the elements.

TABLE III.

55

Svi

32

Xeo 130

Agi 107

75 Crvi

52

204

Tav 181

Biv

207

WVI 184

555

79 Fevi 56

Asv Sevi 78 Mnvi Brvii In each of these 17 pairs of elements there is a difference of 23 in the atomic weights, and at the same time a difference of one valence. Considering any one of the pairs, e.g., As and Cr, it is evident that if a mass of 23 with one valence were added to a mass of 52 with six valences, and if two of the seven became quiescent,( the result would be a mass of 75 with five valences. Similar remarks apply to all the other 16 pairs. TABLE IV.

CHEMICAL NEWS October 1, 1920

merely because there are so many differences of 23, but because the addition of a mass of 23 always corresponds with the gain or loss of one valence. Proof that the above generalisations are not due to chance.

Table III. gives all the cases in which an addition of a mass of 23 to an element produces an element whose valency is either one greater or one less than that of the lighter element; but in order to prove that the state of affairs is not due to chance it is necessary to consider all the pairs of elements whose masses differ by 23, whether the results are favourable or unfavourable. So in addition to the data given in Table III. the following must be considered :-S (32) ... Gl (9), Cl (35) ... C (12), Kr (82) ... Co (59), Kr (82) Ni (59), Nt (222) Hg (199).

...

(a) There are now altogether 22 pairs of elements, each pair differing by a mass of 23. The probability that the difference of the valences of any pair would accidentally be is 1: 2, (where + means that the larger of the pair has one more valence than the other, and -1 means that the lighter has one more valence than the other). E.g., if the valency of the lighter mass were 2, the valency of the heavier might be 1, 3, 5, or 7; or, if the valency of the lighter were 3, that of the heavier might be o, 2, 4, or 6. Therefore the probability that 17 out of 22 would accidentally have a difference in their valences of

in preference to the other two possibilities is I: 222-5×2 = I : 212 = I : 4096.

...

(b) There are ten pairs of elements, given in Table II., with a difference of 46 (=2× 23) in their atomic weights, and no difference in their valences. Five other pairs have a difference in their valences, viz., Mn (55) ... Gl (9), Zn (66) ... Ne (20), Ga (69) Na (23), Te (128) ... Kr (82), La (139) ... Cb (93). The probability that any pair would accidentally have no difference in its valences is 1 : 4. E.g., if the valency of the lighter element had been 4, the heavier might have been o, 2, 4, or 6; and if the valency of the lighter had been 5 the heavier might have been I, 3, 5, or 7. Therefore the probability that 10 pairs out of 15 would accidentally have no difference in their valences is 1: 415-5×11 : 48.

(c) There are nine pairs given in Table IV. with a difference of 69 (=3×23) in their atomic weights, and with a difference of +3 in their valences. There are four other pairs which do not conform to this general rule with regard to their valences, viz., Rb (85) ... O (16), Sr (88) F (19), Ru (101) S (32), Hg (199) ... Xe (130). The probability that any pair would accidentally have a difference of +3 in its valences' is 1: 4. E.g., if the valency of the lighter element were 3, that of the heavier might be o, 2, 4, or 6. Therefore the probability that 9 pairs out of 13 would accidentally have a difference of +3 in their valences is 1: 413-4x=1:47.

(d) Therefore the probability that all three coincidences have happened together by accident is 1 2 12 +16 +14=1: 242 = one to four billion.

(a) proves that it is not a matter of chance that an additional mass of 23 generally produces a gain or loss of one valence; (b) proves that there is no accident in the observation that an additional mass of 46 generally produces no change in the valency (the pair being quiescent); (c) proves that there is no accident in the observation that an

159

additional mass of 69 generally produces a gain of three valences; and the three proofs together (d) make it infinitely probable, i.e., certain that a mass of 23, from which emanates one valence, takes a prominent part in the formation of the elements.

As some may think that this proof is vitiated by the fact that the nearest odd or even integers are employed instead of the actual experimental atomic weights (0=16), the matter may be considered in the following way.

When the experimental atomic weights as given at present are employed, the approximate number 23 occurs 100 times in Tables II., III., and IV., and in 90 of these it lies between 22.5 and 235, and in 50 it lies between 229 and 231, and it is much more extraordinary that one valence can be shown to emanate 100 times from a mass which is always approximately 23 than that one valence can be shown to emanate 100 times from a mass which is always exactly 23.

When a state of affairs has been proved to be due to cause and not to chance it is evident that no future discoveries or additions to knowledge (such as isotopes, e.g.) can ever contradict the general truth obtained, for it is impossible for scientists to unconsciously evolve a state of affairs which can be proved to be due to cause and not to chance, and yet for it not to contain an absolutely incontrovertible truth.

Table II. explains why the principle of the inter-relationship of the elements does not always apply equally well to the heavier elements as it does to the simpler ones. The chief reasons are, first, that more than two of the simpler elements come into the constitution of the heavier ones; secondly, that there are no elements of masses 46 (=2×23) and 92 (=4× 23).

The only numbers which have previously been observed to connect similar elements are 16 and 20; but both these are chiefly due to the number 23 as shown in the CHEMICAL NEWS, December 26, 1919; for the 16 is explained by the replacement of a 7 by a 23, and the 20 is due to replacement of a 3 by a 23. E.g. :

C (12)=Li_H_H,H
Si (28) Na-H-H,-H
P (31) Na-H-H,-H-H,
V (51) Na-H-Na-H-H1.

essential constituent. At least they alone are common to all the acids possessing this quality. The ionic theory may account for the "salty taste" of sodium chloride, sodium sulphate, and other sodium salts in a similar way. The peculiar taste of typically basic substances such as potassium hydroxide and sodium carbonate, is associated with the presence of the common hydroxyl ion. In view of these facts and inferences it was logical to extend the analysis to organic compounds and to try to discover the particular atoms or groups common to substances having a sweet or bitter taste.

re

It has been pointed out by Nef that most of the compounds corresponding to the formula (CH2O)n are sweet. Through the work of Emil Fischer ("Untersuchungen über Aminosaüren," etc., 1906) we became acquainted with the fact that many of the a-amino acids have a sweet taste. L. Henry (Compt. rend., 1895, cxxi., 213) sought to connect the taste of certain bitter compounds with the group -CNO,CH,OH. Finally, G. Cohn ("Die Organischen Gesmachstoffe," 1914) in his markable study of organic tastestuffs compiled an enormous amount of evidence, which proved that there is a close relation between constitution and taste. His work confirmed Henry's theory which he amplified by showing that taste in general is dependent on the presence of certain groups, such as the hydroxyl and amino groups. He calls them sapophoric groups. He points out that those sapophoric groups occur frequently in pairs. Finally, he cites a great number of instances showing that the lower members of certain homologous series have a sweet taste, while the higher members of the same series may be tasteless or bitter. He fails, however, to supply us with a simple theory which could include these facts and which would permit chemists to classify tastestuffs. And we still lack simple rules by which it would be possible to predict the taste of a given compound by a simple inspection of its formula.

Mode of Procedure.

We decided to start our investigation with the sweet aliphatic compounds. They are, of all the tastestuffs, undoubtedly the most important, both from a chemical and physiological point of view. A great number of facts are known in connection with this subject. The relations within each class (sugars, amino-acids, halogen compounds) are pretty well known, while the lack of a theory embracing all the more important among them has been felt most keenly.

In approaching this problem we took advantage of the experience of other chemists who had attempted and finally succeeded in finding a relation between constitution and the dyeing properties of organic compounds. This relationship, as is well known, has been satisfactorily explained by O. N. Witt (Ber., 1876, ix., 522) and others by attributing these properties to two different kinds of groups, the so-called chromophoric groups, that are characterised by a double bond, and the auxochromic groups. A substance containing a chromophoric group is called a chromogene. A chromogene is a coloured substance, a potential dyestuff, so to speak, yet it is only by the introduction of one or several auxochromic groups that those compounds are transformed into real dyestuffs. We shall see that something similar holds true for the constitution of tastestuffs.

The Glucophores of Aliphatic Compounds. As has been pointed out repeatedly by Cohn (Loc. cit., p. 118), the sweet taste of organic compounds is frequently bound to one or two sapophoric groups, which very often occur in pairs. This observation we considered to be a good starting point, but it had to be extended.

Instead of attributing the sweet taste of a given compound to one factor, the glucogene (Cohn) or glucophor, as we prefer to call it, we maintain that it is due to two distinct factors, a glucophore and an auxogluc.

Any glucophore will form a sweet compound with any auxogluc. It is the purpose of the present study to determine just what aliphatic groups may act as glucophores, and which as auxoglucs.

Definition. We define the glucophore as a group of atoms which has the power to form sweet compounds by uniting with a number of otherwise tasteless atoms or radicals.

A preliminary examination of the literature revealed the existence of the following glucophores

-

1. CH2OH.CHOH-. Glycol, CH,OH.CH,OH, which is the most simple substance in which this glucophore occurs, may be considered as a combination of the group CH2OH.CHOH-and hydrogen. Glycol is sweet, as A. Wurtz pointed out

many years ago.

If the same glucophore, CH2OH.CHOH—, is linked to the methyl radical or ethyl radical, we obtain 1,2-propanediol CH,.CHOH.CH2OH, and 1,2-butanediol CH,CH,CHOH.CH2OH, respectively, two other sweet compounds.

If we combine this glucophore with the methylolradical, CH2OH-, then we get glycerin, which is notoriously sweet.

The group CH2OH.CHOH- falls clearly within our definition.

2. CO.CHOH—(H). This second glucophore yields with two hydrogen atoms glycolaldehyde, the simplest sugar.

With the methylol radical CH,OH―, the glucophor -CO.CHOH—(H), yields two different compounds. If the radical -CH2OH is linked to the carbinol of the glucophore, and the hydrogen atom H is bound to the carbonyl group of the glucophore, we obtain the glyceric aldehyde CH2OH.CHOH.CHO.

If furthermore the methylol radical is linked to the carbonyl of the glucophore, while the hydrogen (H) is linked to the carbinol, then dihydroxyacetone, CH2OH.CHOH.CH,OH, results.

All the three compounds named are known to be sweet. The group -CO.CHOH- (H) is then a glucophore. The hydrogen (H) simply indicates that the group must be united with one atom of hydrogen at least, in order to become a glucophore.

3. CO,H.CHNH,–. From the third glucophore most of the sweet amino acids may be derived in a similar way. We know aminoacetic acid, CH,NH,.CO,H, and a-alanine, CH..CHNH,. CO2H, to be sweet. Even higher a-amino acids, such as dl-leucin are sweetish. The group CO,H.CHNH, is therefore a glucophore.

It

4. CH2ONO,-. This is the first example of a glucophore containing but one carbon atom. has been pointed out by Cohn (Loc. cit., p. 411) that many compounds are known which evidently owe their taste to this group. Ethyl nitrate,