POLYNESIAN RESEARCHES. 
511 
They did not enclose their temporary encampments in 
the open fields but each party considered a fortification 
as a security against invasion, and a refuge after defeat 
in the field. 
Their places of defence were rocky fortresses improved 
by art—narrow defiles or valleys sheltered by projecting 
eminences—^passes among the mountains, difficult of ac¬ 
cess, yet allowing their inmates a secure and extensive 
range, and an unobstructed passage to some spring or 
stream. The celebrated Pare, in Atehuru, was of this 
kind; the mouth of the valley in which it was situated 
was built up with a stone wall, and those who fled thither 
for shelter, were generally able to repel their assailants. 
Sometimes they cut down trees, and built a kind of 
stage or platform called projecting over an avenue 
leading to the pare; upon this they collected piles of stone 
and fragments of rock, which they hurled down on those 
by whom they were attacked. In some of the Harvey 
Islands they planted trees around their places of encamp¬ 
ment, and thus rendered them secure against surprise.— 
These enclosures they called pa^ the term which is used 
to designate a fort in the Sandwich Islands. 
If those who had been routed on the field of battle were 
allowed by their pursuers time to wall up the entrances 
of their places of refuge, they were seldom exposed to 
assault, though they might be decoyed from them by 
stratagem, or induced to leave from hunger. The pan 
in Borabora, and some places in Tahiti, are seldom ex¬ 
celled as natural fortresses. Several of these places were 
very extensive : that at Maeva, in Huahine, bordering on a 
lake of the same name, and near Mouna-tahu, is probably 
the best artificial fortification in the islands. It encloses 
many acres of ground well stocked with bread-fruit, con- 
