r 
Melville Bay.] TERRA AUSTRALIS. 
of Carpentaria; the entrance is from the N. N. W., four miles wide, 
and free from danger ; and within side, the sole dangers not con- 
spicuous, are a sandy spit running half a mile to the S. S. E. from 
Point Dundas, and the Half-tide Rock. This lies half a mile from 
the north-west part of Drimmie Head, and bears (true as usual), 
From the sandy hillock within Point Dundas, - S. 48° 35' E. 
From Harbour Rock, - - - S. 10 39 E. 
Melville Bay every where affords good holding ground, the 
bottom being either mud or sand ; and there is depth for a ship to 
run between the three rocks in the middle of the bay and Drimmie 
Head, and steer eastward until the head is brought to bear N. N. W., 
at the distance of one or two miles; but the most convenient anchor- 
age is just within the entrance, between Point Dundas and Harbour 
Rock, where a ship may lie close to the sandy beach in from 3 to 5 
fathoms. Even within the rock there is depth enough ; and were 
moorings laid down, four or five sail might swing there in perfect 
security. We obtained here fire wood, and a tolerable supply of 
fish ; and had water been wanted, it might have been obtained by 
digging at the foot of the small hills to the north-east of Harbour 
Rock, since a hole made there by the natives was found to contain 
good water. 
The stone on the north side of Melville Bay is a granitic com- 
position of quartz, mica, and coarse garnets ; the garnets are large, 
and give the stone a plum-pudding-like appearance, and when 
polished, it would be beautiful : over the granite is a crust of cal- 
careous rock in many places. On the south side of the bay the stone 
is argillaceous, but frequently mixed with ferruginous grains ; and 
on the south-east side the rocks are of iron ore, of which a small 
piece drew the needle of my theodolite 8° from the meridian. The 
bearings taken here were found to have been 50° wrong ; but too late 
to ascertain whether the error arose from the attraction of the shore, 
or from the needle having been placed at 310° by mistake, instead 
of 360°. 
Gg 
225 
1803. 
February. 
VOL. II. 
