30 Primitive Architecture. 
arrange the fires in the central aisle, so that one fire serves for four 
cabins. More developed are the dwellings of New Guinea, with a. 
fire to every two cabins, and of the Mishmis, with a fire to each 
cabin. Another form is found among the Mayas, who build a sep- 
arate cook-house where the cooking for the whole village is done. 
The Ostyaks keep their food safe from the dogs in a village store- 
There is no more singular mode of building than that of elevat- 
ing the dwelling on poles. It is of most frequent occurrence among 
communistic peoples, but is by no means confined to them. Its 
origin has been long a favorite subject for controversy among stu- 
dents of primitive architecture. The historians of Timor allege 
that it arises from the fear of the reptiles that infest that fertile 
island, and we are also informed that such houses are constructed at. 
Kurrecchane that the children may sleep safely at night. HoAvever 
well this custom in these places may be explained by these state- 
ments, it is sufficiently obvious that the explanation is not a universal 
one, and its origin must be looked for elsewhere. The best theory 
yet proposed is that of M. Frederick Troyon,^ but which, though it 
is supported by many facts, fails when put to the test of universality. 
Beginning with the observation that all such buildings are built 
over or near water, M. Troyon argues that the rafts used in the 
early migrations afforded little protection to their owners, especially 
when the men were off hunting. Safety, however, was readily ob- 
tained by mooring in midstream, while, when pulled ashore, the raft 
was best kept from being washed away by the waves, by being 
elevated beyond their reach. Unfortunately for his theory, how- 
ever, M. Troyon has ignored the fact that elevated houses are to be 
found both on the coast line and in interior districts Avhere rafts 
would be impractical. Other and possibly many causes have 
contributed to the custom ; among- them especially the desire for 
greater protection. It is not sufficient for the Sumatrans'^ that 
they hide their dwellings amid the trees on a hill-top, to which 
there is but one, or at most two, narrow paths of access, nor is a 
high and strong fence enough. They elevate their houses on posts 
and enter by means of movable notched poles. The theory of pro- 
tection is confirmed by the solitary houses being more elevated than 
are the village houses. If the custom of building elevated houses 
' Troyon, Habitations Lacustres des temps anciens et raodernes. Lau- 
sanne : 1860. 
* Marsden, Historylof Sumatra, 56. 
