Mr. M. M. Piittitiou Muir on Cheiulcal Clast<ijication. 95 



cepted numbers are the true atomic weights is very great ; it 

 never, however, amounts to certainty. But in all our measure- 

 ments we can only deal with greater or lesser degrees of probabi- 

 lity, and with probability we must be content. The commonly 

 accepted formula for potassium chloride is KCl, the atomic 

 weight of chlorine is almost certainly 35*37, the atomic weight 

 of potassium is very probably 3y*13; but the kind of evidence 

 upon which we base the assumption that the molecular weight 

 of potassium chloride is represented by the formula KCl 

 would lead us to the formula CH as expressive of the mole- 

 cular weight of benzene. Conclusions as to valency drawn 

 from a consideration of compounds whose molecular weights 

 are unknown (that is, from compounds the densities of whose 

 vapours have not been determined) are more or less misleading. 



14. The second diihculty which must be overcome before 

 the valency of an element can be known is the determination 

 of the atomic weight of that element. The maximum atomic 

 weight is best determined by estimating the smallest relative 

 quantity of the given element contained in two volumes of any 

 of its gaseous compounds, the usual units being employed* 

 Here, again, we are obliged to deal only with gaseous com- 

 pounds. Failing this method, or as a check upon the deter- 

 minations made by this method, we ha^ e the methods of specific 

 heat and isomorphism, and, lastly (at presentmost indeterminate 

 of all), the method of general chemical analogies. Although 

 the first method is undoubtedly the best, it may in certain 

 cases be advisable to adopt the result obtained by methods 2, 

 3, and 4 in preference to that obtained by method 1. The 

 commonly accepted atomic weight of iron is a case in point* 



15. Besides these two main difficulties, viz. determinations 

 of molecular and atomic weights, there ai^e other subsidiary 

 difficulties attending the estimation of the valency of an ele- 

 ment. Chief among these stand the difficulties arising from 

 the existence of so-called '^ unsaturated compounds " and of 

 so-called " molecular compounds." 



16. Compounds are known in which the whole of the '' equi- 

 valents " or '' combining-powers " of one of the constituent 

 atoms are not saturated. Thus, although the great mass of 

 evidence obliges us to regard nitrogen as a triad, or more 

 probably as a pentad, and oxygen as a dyad element, the 

 compound NO certainly exists. In this compound one at least 

 of the " combining-poM^ers " of nitrogen must remain unsatu- 

 rated. So also the compound CO is an unsaturated compound, 

 inasmuch as we must regard the carbon atom as tetravalent. 

 These unsaturated compounds usually combine, with tolerable 

 ease, with additional atoms: CO readilv forms CO CL &c. 



