sample (2 ha.) for each volcano show an increase with age of the blast area and are all far below the numbers found in a comparable stage of garden regrowth. It is apparent then that the great majority of tree species originally present have been destroyed and that fresh species are gradually entering the flora of the blast areas. These species increase in number and proportion on the older blast areas where they are mainly found as rare individuals in sub-dominant layers of the communities. The entry of these species must be very gradual as even after 80 years the tree flora of Mt. Victory is still far poorer than a similar extent of climax forest. This increase in number of species is similar to the steady increase on Krakatau where investiga- tions have been far more extensive (Treub, 1888; Docters van Leeuwen, 1936).
DISCUSSION The various communities described for each individual volcano show marked cor- :relationswith the environment. On the Waiowa Volcano, where the climatic range is small, there is a direct correlation with edaphic conditions. The recent communities on Mt. Lamington also show a major response to edaphic factors with only slight modifications due to climatic factors. Climatic factors, however, exert the greatest influence on vegetation types now present on Mt. Victory. This would indicate that as the succession progressed, the amelioration of edaphic conditions would make climatic variation of greater significance.
The actual course of the various successions cannot be traced but may be inferred by a comparison of communities present in the three volcanoes. On Mt. Victory lahar deposits carry a forest community which is similar in structure to stages in garden regrowth, but similar deposits on Mt. Lamington and Waiowa Volcano now carry an open herbaceous community. These open communities can therefore be expected to be dominated by trees and a closed canopy to be formed within the next 50 years. Similarly the dome of Mt. Lamington, the most unfavourable site for plant growth within the blast area, is now devoid of vegetation, yet the dome of Mt. Victory, apparently of very similar structure, can be seen from air photographs to be covered by a closed woodland community. This is in accord with successions traced on Krakatau where coarse deposits bearing an open fern community in 1886 carried a woodland community in 1932 and the succession is apparently heading towards climax rainforest (Richards, 1952).
Unlike Krakatau, there are many sites within the blast areas which have been pioneered by tree species rapidly forming a closed woodland community. Such communities occur whenever edaphic conditions are not too severe, and the future course of such successions will depend on the height and age to which the pioneer species can grow and the rate of entry of other species into the community. The Octomeles sumatrana community on Mt. Victory is dominated by a large number of light-demanding species which could only have become established in the early stages of the succession. A percentage of these trees is dying, leaving a break in the canopy which is being filled up by species present in the second and third layer, a proportion of which are species characteristic of climax forest. I t is expected, however, that the change over to a climax forest will take several centuries to complete.