( 420 ) 
III. Dichloroacetyl chloride, on being heated gently, is also decora- 
posed in two directions. 
A CCljHCOCl = CO + CC1„H ± 60° /o 
B CClgHCOCi = CO + HCI + CC1 S 
the latter being presumably polymerised to C S C1 10 . 
IV. Trichloroacethyl chloride, when distilled repeatedly with A1CI, 
is decomposed in one direction: 
CC1 S C0C1 = CC1 4 -f- CO 
The first part of the research, a contribution to the theory of the 
decomposing action of aluminium chloride was herewith attained. 
II. 
As regards our second object, namely, the finding of the point 
attacked in the decomposition, the different behaviour of chloral and 
dichloroacethyl chloride gives already some indications. 
If we assume that one of the Cl-atoms of the chloral is detached 
we may expect it to combine with the movable H-atom to form HCI 
and this^ indeed takes place. If the CO group or the H-atom were 
the point attacked we ought to obtain chloroform which I have now 
been able to isolate in very small quantities only. 
On the other hand, and for the same reason, the formation of 
chloroform from dichloroacethyl chloride is to be expected; it is now 
the movable chlorine atom which is detached first; CO and CC1,H 
then follow immediately. Being the most unsaturated one the latter 
group will combine with the chlorine atom to chloroform, hydrogen 
chloride occurring as by-product. The actual course of the reaction 
is therefore quite in harmony with the assumption that primarily 
the movable chlorine atom is gendered active. 
The action of aluminium chloride on trimethylacetyl chloride is too 
violent to decide anything as to the first stadium of the reaction. 
[That the chlorine atom exercises again a preponderating influence 
on the instability of the molecule is shown from the behaviour of 
the otherwise quite analogous pinacolin; this simply unites with the 
catalyst without forming a trace of CO even at 100°]. 
I have, therefore, observed the action of concentrated H,S0 4 w 
regard to trimethylacetyl chloride because the behaviour of this aci 
is in many cases analogous to that of A1C1, and also because it 
forms the same kind of additive products (Recueil 1905, p- 222)- 
