474 M. Fromme on the Changes produced in the Molecular 



vations which I had published ; on the contrary, their result 

 had been so remarkable and unexpected that further experi- 

 ments were much to be desired. Since I have not in the mean 

 time become acquainted with investigations on this subject, I 

 offer now a small contribution to the solution of the question. 



1. On the Change in Density produced by Tempering Iron, 



The following materials were employed : — 



(a) Iron wire 1*4 millim. thick. 



(b) Iron wire 0*5 millim. thick, obtained by drawing out 



the wire (a). 



(c) Iron wire 0*3 millim. thick. 



(d) Steel wire 0'7 millim. thick. 



(e) Steel wire 0'4 millim. thick, (d) and (e) were from 



the same works in Birmingham, and were called 

 " patent steel music-wire." 

 (/) Cylindrical steel rod 5 '5 millim. thick, 30 millim. long. 

 (g) Electrolytic iron in small pieces, deposited from a solu- 

 tion of ferrous sulphate. Thickness about Ol millim. 

 (A) Electrolytic iron deposited from a solution of ferric 

 chloride. Three small pieces were cut from a large 

 piece and tested separately, Mean thickness of the 

 large piece 0'14 millim. 

 (i) Cast iron in the form of a cylinder, 18 millim. long and 



7 millim. thick. 

 (&) Cast iron in the form of a plate (20 : 10 : 5 millim.). 

 (I) Wrought iron in the form of a cylinder 25 millim. long 



and 5' 6 millim. thick. 



(m) White pig-iron in small pieces, such as could be obtained 



from a larger piece by breaking it up with a hammer. 



For determination of density, a Geissler's pycnometer was 



employed, except for (h) and (&), the densities of which were 



determined with the hydrostatic balance. The metal was heated 



in the flame of the Bunsen burner or in the blowpipe-flame. 



When slow cooling was desired the pieces of metal were 

 allowed to cool in the bright flame and above it, whilst rapid 

 cooling was attained by plunging the hot body into water of 

 10° or 15°. The coating of oxide was removed before each 

 determination. The wires were examined by cutting off pieces 

 suitable to the height of the vessel of the pycnometer, of about 

 4 centim. long, and testing a number together (thirty in the 

 case of the thinnest wires). 



Each specimen was first heated and cooled slowly, and then 

 its density Si measured ; then it was heated again and cooled 

 rapidly, and its density B r determined. Thus determinations 

 of S, and 8 r alternated with each other, often repeatedly. 

 Table I. exhibits all the observations brought together. 



