OF ARTS AND SCIENCES. 315 



which had floated in to the distance of from ten to one hundred 

 millimeters from the crust, and on the inner edge of these cloudy por- 

 tions the crystals became more distinct, and finally shot out in crystal- 

 line plates (which were seen of course only in section) some of them 

 from fifty to one hundred millimeters in extent. These again were 

 met and intersected by similar crystalline plates radiating from numer- 

 ous centres through the mass, these centres being spots of more com- 

 pact crystallization, like the cloudy masses around the edge, forming in 

 the interior a network of plates producing skeleton crystals. 



Thus the entire surface presented exactly the appearance that one 

 might expect in a mass of iron which had slowly cooled. In such a 

 mass, the surface cooling first would become compact, and exhibit only 

 confused crystallization, while the interior, cooling more slowly, would 

 allow time for isolated crystal plates to form ; and at the same time 

 local causes would start a simultaneous crystallization from other 

 points in the interior, determined probably by the presence of nuclei 

 of foreign matter. 



The possible variation in the character of the figures brought out by 

 etching different portions of the same iron, has been mentioned in a 

 former paper,* but at that time no such striking example had been 

 noticed as that now exhibited by the Coahuila irons. Specimens 

 might be cut from the iron shown in Plate I. so selected that one piece 

 when etched would give well marked Widmanstattian figures, with all 

 the characteristic features which have commonly been associated with 

 the so called typical octahedral irons, while another would show 

 equally typical Neumann lines, and still a third piece would appear 

 perfectly amorphous or made up of irregular grains. 



WlDlIANSTATTIAX FiGUKES ON " .SPIEGEL ElSEN." 



Hitherto Widmanstattian figures and Neumann lines have been 

 considered the strongest characteristic of meteoric irons ; but since it 

 is merely the evidence of slow crystallization attended by the exclu- 

 sion of foreign matter, one would expect that any impure iron, pro- 

 vided that it had cooled slowly from a state of fusion, would exhibit 

 similar characters. From depending too much upon these features the 

 Greenland iron was originally considered to be of meteoric origin, as 

 some of the masses show very good Widmanstattian figures. Since 

 specimens of cast iron have been frequently described as meteoric, and 



* Proceedings of the American Academy, vol. xxi. p. 478, May, 1886. 



