In the compounds of bromine and chlorine (the latter more particularly) the divisor is too much below its normal value, possibly because the atomic numbers of these compounds have been determined from their specific gravity in the liquid state and may therefore be too small. It is also worthy of remark that the divisor always has its smallest values in the combinations of potassium and sodium with iodine, bromine and chlorine; then follows lead, then silver and then mercury. It is possible according to this view to calculate the hypothetical atomic number of oxygen from its combinations; but different compounds give widely different results, the numbers thus found varying from 1900 to 4100; and since, whatever number is assumed as the true one, the oxygen compounds exhibit relations differing from one another much more widely than those above considered, it seems better to leave this hypothetical part of the calculation on one side till the causes which may affect the magnitude of the divisor shall have been better ascertained. The great variation in the results will be seen from the following examples. Since according to the law under consideration the atomic number of a metallic oxide should be found by adding the atomic number of the metal to the atomic number of oxygen and dividing the sum by 4, (the oxide being supposed to contain 1 At. metal and 1 At. oxygen), it follows that if we multiply the atomic number of such a metallic oxide by 4 and subtract the atomic number of the metal from the product, the remainder will be the atomic number of oxygen. By treating different metallic oxides in this manner we obtain the following results: Potash; (625.4) — 245 = 2251. Soda; (999. 4) — 466 = 3520. Oxide of zinc; (1548.4) = 3806. Oxide of copper; (1795. 4) — 3046 = 4136. Suboxide of copper Cu2 O, in which there are 3 atoms and consequently the divisor = 9, gives (892. 9) — 2.3046 = 1936. = 2386 The same method applied to compounds of the second order gives the following results: = Anhydrous Carbonates. Ba O, CO2 (687 + 419): 2·3 = 481 (484). Sr O, CO2 23 547 (543).-Ca O, CO (Arragonite): 2.5 660 (664). PbO,C02:25=533(533)—Zn0,CO:25=787(781...790).—Ag 0, CO2 : 2.5 = 484 (489).-KO, CO2: 2.8 = 373 (363).—Na O, CO2: 2-8 = 507 (514).—Ca O, CO2 (Calcspar): 28 = 590 (599).—Mg O, CO2: 2.8 = 766 (797).-Mn O, CO2: 2·8 = 673 (684).—Here the divisor varies only from 2.3 to 2.8: that it is not equal to 4, may perhaps arise from the atomic number of carbonic acid having been calculated from its specific gravity in the liquid state: there can be no doubt that the sp. gr. and therefore also the atomic number of solid carbonic acid is much greater. Anhydrous Sulphates. Ba O, S 03 (687 +547): 31388 (400... 423). Sr O, SO3: 32 = 433 (433).-Pb O, S O3: 3·3 = 451 (451).—Ca O, SO: 3.5479 (474).-AgO, SO3: 3.5 = 382 (380).-K O, S 03: 3.5= 337 (334).-Na O, SO3: 37 418 (410).-Zn O, SO3: 4.5= 473 (471). Mg Ò, SO: 47 482 (477).-Cu O, SO3: 47498 (498). Anhydrous Nitrates, the atomic number of nitric acid being hypothetically assumed 300. Ba O, NO (687 + 300): 37267 (271). Sr O, N 05:39 293 (294).-AgO, NO: 3·9 279 (284).-KÓ, NO: 4231 (225....230).—Pb Ŏ, NO3: 4·2 = 300 (294).—Na O, N 05: 4.4 295 (290). In all these three groups, the salts of baryta and strontia require the smallest divisors; then follow oxide of silver and potash, then soda, then magnesia; the position of lime, oxide of lead, and oxide of zinc is variable. Hydrates. KO, HO (625 +1234): 4.5 = 413 (414).-Na O, HO: 4558 (552).—S O3, HO: 4.2 424 (418).-S O3, 2HO: 12 =251 (252).—SO3, 3H O: 16265 (270). These cases are nearly in accordance with the law; the divisor of SO3, 2H O should however be 9 instead of 12. The first hydrate of nitric acid does not give so close an approximation: NO, HO (300 + 1234): 57269 (268). For gypsum we have Ca O, SO3 + 2H Ò (474 + 2.1234): 10 294 (298), the divisor being 10 instead of 9. The law thus developed may be expressed by the following formula: x Z' + yZ" =Z" in which Z' denotes the atomic number of the first element, Z" that of the second, Z" that of the compound, x the number of atoms of the first element, y that of the second, which make up the compound atom. In the application of this law, striking deviations are apparent, as may be seen from the foregoing calculations. The most remarkable exception is presented by sulphuret of carbon, CS; (6481 +2.1388): 25369 (372); here the divisor is 25 instead of 9, as if the number of atoms in the compound were 5 instead of 3. Even allowing that many exceptions may arise, partly from errors in the supposed atomic weights of the simple substances, partly from incorrect determinations of the specific gravities of these simple substances and their compounds, the law must still be looked upon as far from the truth; it must needs be modified by many circumstances of much greater importance than those just noticed, and not till these circumstances have been discovered and reduced to calculation, shall we be able to regard the law as established on a proper foundation. The different degrees of cohesion in compounds and their elements, and their different attraction for heat would probably be found of particular importance. It is to be wished that this matter could be subjected to careful mathematical analysis: the foregoing table might furnish the required data. The study of the relation between atomic weight and density has received some important additions since the publication of the last edition of this work. Of the researches of Kopp, Schröder, Löwig, and others of the continental chemists an elaborate account is given in the report of Professor Otto, a translation of which appears in the volume of Reports and Memoirs lately brought out by the Cavendish Society. The recent memoir of M. Filhol (N. Ann. Chim. Phys., 21, 415), included in the same volume, likewise gives a review of the labours of previous experimenters in this branch of science, and adds some interesting results obtained by the author himself. I shall here notice some of the leading points in M. Filhol's memoir. The author's calculations are founded on the following atomic weights and densities, the latter determined by his own experiments. Bi-oxide of copper (black oxide).. 6.322 Hydrate of potash 5.612 2.240 .. 9.361 4.581 2.044 1.994 soda 2.130 sodium 2.240 .... baryta (Ba O, HO).. 4:495 2.960 baryta (Ba O, 9H Ó) 1.656 .... 2.240 iron (Fe Cl) 2.528 Most of these densities agree very well with those formerly determined by Boullay, Mohs, Karsten, &c., but there are some, e.g., those of baryta and strontia, which differ considerably from former determinations, as will be seen by reference to the table pp. 68....72. Now it has been shown by Kopp that all the metals enter into combination without change of volume, with the exception of potassium, sodium, barium, strontium, magnesium, calcium and aluminum. With regard to barium and strontium, it must be observed that the volumes which Kopp assigned to these metals were deduced from the specific gravities of baryta and strontia as determined by Karsten. Filhol however shews that by adopting the densities given in the preceding table, barium and strontium may be removed from the list of exceptions. Thus the density of baryta being 5.456, its equivalent volume is 254 175, and estimating the volume of oxygen in these oxides at 32 (Kopp, N. Ann. Chim. Phys. 4,462) we have for the equivalent volume of barium, 175 - 32 143. Similarly for strontium, the equivalent volume of strontia being 64560 = 140, the equivalent volume of the metal is 140 32 108: and these are the volumes with which, according to Kopp, the metals barium and strontium enter into other combinations. Schröder has remarked that if from the equivalent volumes of a series of analogous compounds (oxides, chlorides, &c.) the volume of one of the elements be deducted, a constant remainder is often obtained for the element common to the whole series. On applying this process to nitrate of silver and nitrate of lead, and deducting the equivalent volumes of the metals contained in these salts, a constant remainder 358 is left for the radical NO. If on the other hand the equivalent volumes of the oxides be deducted, the silver salt gives for N 5 the remainder 294, and the lead salt 326. Kopp regards this as an argument in favour of the theory of hydrogen-acids. Filhol however shows by numerous examples that with regard to a considerable number of salts, the subtraction of the volume of the base leaves a constant residue for the oxygen acid. Thus sulphuric acid appears to have two different volumes in combination: the sulphates of lead, baryta, potash, and strontia give for SO3 the remainder 154, those of copper, zinc, lime, magnesia, and soda give 204.-The chromates of lead and potash give for Cr O' the remainder 196.-The tungstates of lead and lime give W 03 212.-The carbonates of cadmium, iron, manganese, lead, zinc, baryta, lime, magnesia, potash, soda, strontia, and and the double carbonate of lime and magnesia, give for carbonic acid the volume 119; and lastly, the nitrates of lead, ammonia, potash, soda, baryta and strontia give for N O3 the volume 326. The discrepancy which occurs in the case of nitrate of silver is attributed by Filhol to an error in the determination of the density of oxide of silver. On the whole therefore, the results of these calculations cannot be regarded as decidedly favourable to either theory of the constitution of salts in preference to the other. Filhol likewise points out some remarkable relations with regard to the changes of volume which take place in combination. Thus with regard to the chlorides: let P denote the atomic weight of chlorine, d its density; P' the atomic weight of the metal, d' its density; and ▲ the density of the chloride on the supposition that the elements unite with(P+ P') dd' out change of volume: then A= Pd+P'd If now D denote the density of the chloride determined by experiment, D. Δ we always find D > A, and D expresses the coefficient of con traction. The same method is applicable to other compounds (sulphates, carbonates, &c). The following table contains a series of results obtained in this manner. The density of chlorine is estimated at 1:35, being the mean of the results obtained by Faraday and Karsten, viz. 1.33 and 1.38; that of carbonic acid at 0.83. It will be seen from this table that in a very large number of cases, independently of all hypothesis, the coefficients of contraction of those compounds which most strongly resemble one another in a chemical point of view, are sensibly the same. It may likewise be observed that in a certain number of cases the ratio between the calculated densities of certain compounds of the same kind is the same as that which exists between the densities determined by experiment. Hence we may calculate, à priori, the densities of certain salts according to those of their elements and that of a salt of the same species |