438 Prof. Tyndall on the Action of Free Molecules on 



blished with a clearness and an evidence unattainable by any 

 other means then existent *. 



Notices of the investigation having appeared in many 

 English and continental journals f, I was induced to defer the 

 detailed publication of the experiments. The investigation 

 itself had taught me the difficulties and dangers which beset 

 it. These had reference both to the methods of experiment 

 and to the purity of the substances employed. To secure the 

 perfect constancy of the sources of heat, and the perfect 

 steadiness of the galvanometer, when the flux of heat was 

 powerful, involved a lengthened discipline. With neither 

 gases nor vapours, moreover, was it easy to obtain uniform 

 results. When generated in different ways, the action of the 

 same gas would sometimes prove itself so discordant as to 

 suggest to me the possible existence of novel allotropic con- 

 ditions to account for such variations of behaviour. Two 

 samples, moreover, of nominally the same liquid, would furnish 

 vapours yielding results far too divergent to be tolerated. 

 The drying apparatus also contributed its quota of disturbance. 

 These anomalies were finally traced to the fact that an in- 

 credibly small amount of impurity, derived from the stronger 

 gases or vapours, sufficed to disguise and falsify the action of 

 the weaker ones. All this had to be learnt; and when learnt, 

 1 thought it desirable, for the sake of accuracy, not to publish 

 the results which had been gained with so much labour, but 

 to go once more, with improved appliances, over the same 

 ground. This I did, though it involved the total abandonment 

 of seven weeks' uninterrupted experimental work in 1859, of 

 seven weeks' similar work in 1860, and of many fragmentary 

 efforts. On the 10th of January, 1861, the memoir con- 

 taining an account of the investigation was handed in to the 

 Royal Society \ . 



* With moderate total heats the method of compensation is extremely 

 easy of application; but when the total radiation is very large, some 

 discipline is required to keep the galvanometer-needle steady in its most 

 sensitive position. With due training, however, perfect mastery over 

 this difficulty may be obtained. 



t Proceedings of the Royal Society, May 26, 1859; Proceedings of the 

 Royal Institution, June 10, 18o9 ; Bibliothhque Umv&'selle, July 1869; 

 Cosmos, vol. xv. p. 321 ; Nuovo Cimento, vol. x. p. 196; Comptes Rendus, 

 1859 ; and in other journals. 



% Section 8 of the Bakerian Lecture for 1861 reveals some of the 

 difficulties which beset the earlier stages of these inquiries. To secure 

 strength of radiation and steadiness of the needle I passed from source 

 to source, obtaining my temperatures in turn from water, oil, fusible 

 metal, sheets of copper heated by regulated flames, and from other things. 

 Approximate results were readily obtainable ; but I aimed at a degree 

 of accuracy which would render any material retractation afterwards 

 unnecessary. Soundness of work I thought preferable to rapidity of 

 publication. 





