1892.] on Metallic Carhomjls. 677 



Although the solution resembles in appearance to some extent 

 the solutions obtained by treating potassium carbonyl with water, 

 it does not give any of the characteristic reactions of the latter. 

 When speaking of potassium carbonyl, I mentioned that by its 

 treatment with water, croconate of potassium was obtained, which 

 has the formula K2C5O5. 



We have transformed this by double decomposition into ferrous 

 croconic, FeCgOg, a salt forming dark crystals of metallic lustre 

 resembling iodine, which is not volatile, and dissolves readily in 

 water, the solution giving all the well-known reaction of iron in 

 croconic acid. You will note how entirely different the properties 

 of this substance are from those of iron carbonyl, which I have 

 described to you ; yet, on reference to its composition, you will find 

 that it contains exactly the same number of atoms of iron, carbon, 

 and oxygen, as the latter. This is a very interesting case of iso- 

 merism, considering that both compounds contain only iron, carbon, 

 and oxygen. The difference in the properties of these two bodies 

 becomes explainable by comparing the structural formula of the two 

 substances. 



I would now call your attention to the great difference in the con- 

 stitution of the potassium carbonyl and that of the nickel and ferro 

 carbonyl. In the former the metal potassium is combined with the 

 oxygen in the carbonyl ; in the latter the metals nickel and iron are 

 combined with the carbon of carbonyl. In the first case we have a 

 benzole ring with its three single and three double bonds ; in the 

 second a closed chain with only single bonds. It is evident that the 

 chemical properties of these substances must be widely different. 



The ferro-penta-carbonyl remains perfectly unchanged in the 

 dark but if it is exposed to sunlight it is transformed into a solid 

 body of remarkably fine appearance, of gold colour and lustre, as 

 shown by the sample in this tube. 



This solid body is not volatile, but on heating it in the absence of 

 air, iron separates out and liquid ferro-carbonyl distils over. If, 

 however, it is heated carefully in a current of carbonic oxide it is 

 reconverted into the ferro-penta-carbonyl and completely volatilised. 

 We have so far found no solvent for this substance, so that we have 

 no means as yet of obtaining it in a perfectly pure state. Several 

 determinations of the iron in different samples of the substance 

 have led to fairly concordant figures, which agree with the formula 

 Fe^iCO)^, or di-ferro-hepta-carbonyl. 



The interesting properties of the substances described have 

 naturally led us " to try," as Lord Kelvin once put it to me so 

 prettily, " to give wings to other heavy metals." We have tried all 

 the well-known and a very large number of the rarer metals ; but 

 with the exception of nickel and iron we have so far been entirely 

 unsuccessful. Even cobalt, which is so very like nickel, has not 

 yielded the smallest trace of a carbonyl. This led me to study the 

 question whether, by means of the action of carbonic oxide, the separa- 



