Condition of Iron by Heating to Redness and Cooling. 483 



water ; whence it follows that they are not produced by the 

 sudden cooling, but are to be regarded as indicating original 

 want of homogeneity in the bar. 



A comparison of S 2 with 8 3 shows that the density of the 

 portion of the bar dissolved away is only twice decidedly 

 greater than that of the remaining portion. There is there- 

 fore no decrease of density from the surface towards the axis 

 in the slowly cooled bar. 



The foregoing experiments, taken together, therefore con- 

 firm the common belief that, in consequence of rapid cooling, 

 a distribution of density results, in which the density decreases 

 from the outside towards the centre. 



I have so far spoken only of the distribution of density in 

 a tempered steel bar. According to the usual view, supported 

 also by Mousson, not only the density, but also the hardness 

 becomes smaller from the outside towards the centre, so that 

 a harder shell encloses a softer kernel. 



During the gradual dissolving of the bars by acid, they 

 were tested with a good file as to their hardness. All four 

 bars showed themselves so hard through the entire substance 

 — they were dissolved away so far that the remainder was only 

 1*5 millim. thick — that they could not be scratched by the 

 file. The slowly cooled bar, on the contrary, was throughout 

 very soft. 



Although we must admit the possibility of the tempered 

 bar being harder outside than inside, yet the assumption of a 

 softer kernel enclosed by a harder shell is not proved, and 

 would probably not be found to be the case in all steel bars 

 of thickness less than those employed in these experiments. 



Such bars therefore, whether cooled in warm or in cold 

 water, must be regarded as homogeneous in respect of hardness. 

 It must remain for further experiments to decide whether 

 this result holds good for still thicker bars, or how thick the 

 bar may be, so that the interior shall remain soft in spite of 

 tempering. 



If we take this result together with that of the experiments 

 upon density, we see that in tempered steel hardness and 

 density do not go together, but that smaller hardness may 

 correspond to greater density, and vice versa. For, according 

 to Table III. A, of seven layers removed from a steel bar four 

 possessed a density considerably smaller than the mean density 

 of the slowly cooled soft bar, and yet were quite hard. The 

 same result is arrived at by observing that the outer layer 

 of the bar (e), according to Table III. B, possessed a calculated 

 density of 7 '93, and yet, in spite of the high density, was 

 perfectly soft. 



