BETWEEN THE VISCOSITY OF LIQUIDS AND THEIR CHEMICAL NATURE. 105 
with calcium chloride. It boiled entirely between 63° and 68°. On treatment with 
sodium it became evident that this product contained a very large quantity of 
undecomposed ethyl iodide, and it was found practically impossible to purify it by 
any means short of further etherification. In a second experiment more alkali was 
used, and in this case the mixture of propyl alcohol and caustic potash was kept hot, 
and ethyl iodide added through the reflux condenser at intervals. Gas was still 
given off, and the product obtained again boiled between the same limits and 
contained much iodide. 
It will be seen that the disadvantage of Cha.ncel’s method is that the large 
quantity of caustic alkali converts much of the alkyl iodide into olefine, while, on 
the other hand, unless a great excess of potash is used, the product contains 
undecomposed iodide. 
Although an alcoholic solution of caustic alkali is stated to contain a large pro¬ 
portion of alcoholate, yet the other method, in which the alkali metal is dissolved in 
the pure alcohol, differs very markedly from Chancel’s, and, as will be seen from 
the experiments presently to be described, is greatly superior as a quantitative 
method for the preparation of ethers. This does not, however, immediately appear 
from the existing records. Beuhl (‘Annalen,’ vol. 200, p. 177) acted upon sodium 
ethoxide in alcoholic solution with propyl bromide ; he found propylene evolved, and 
the yield of ether low. Zander (‘Annalen,’ vol. 214, p. 163) employing sodium 
propylate with propyl iodide, obtained only a 25 per cent, yield of dipropyl ether. By 
the use of solid sodium ethoxide and butyl bromide, Lieben and Bossi obtained 
ethyl butyl ether as chief product, with a small quantity of butylene ; but this 
experiment was performed in a sealed tube and is consequently not quite comparable. 
It is noteworthy, however, that only 1’5 times the theoretical quantity of the 
ethoxide was used, while to decompose the less stable butyl iodide Saytzew {loc. 
cit.) had to use six times the calculated amount of potassium hydroxide in Chancel’s 
process. 
A careful comparison of all the above-mentioned results with others by Balard, 
Guthrie, and Williamson, has led, in the light of experiments described below, to 
conclusions which may be stated as follows ; they refer to the preparation of ethers 
containing alkyl radicles from methyl to amyl inclusive. 
i. The action of a solution of sodium in an anhydrous alcohol upon an alkyl halide 
tends quantitatively to the production of an ether,* and differs markedly from the 
action of an alcoholic solution of an alkaline hydroxide. 
ii. In preparing mixed ethers by this reaction, the halogen-derivative should 
contain the alkyl radicle of lower molecular weight, the reaction then being 
apparently the more simple and regular, and the yield better the greater the diffe¬ 
rence between the two alkyls. 
iii. Where the use of the halogen-derivative of propyl or a higher alkyl becomes 
* Cf. Hecht and Coxead (‘ Zeits. Physik. CLem.,’ 3, 450), on this reaction. 
MDCCCXCVII.—A. P 
