evaporation values is 4.6 litres, while corresponding figures between 2,000ft and 1,000ft are 36in and 5.8 litres. Temperature climates have proved to depend mainly on altitude, and inversion patterns on frosty nights to be marked only in grassland. The distribution of some species appears well-correlated with these climatic values. Most of the lowland forest trees have a well-defined upper limit lying, in the case of rimu, miro and pokaka, between 1,300ft and 1,600ft. The distribution of the montane conifers can be correlated with meteorological data thus:—

The relative distribution of forest and grassland is not, however, readily explained. The great humidity of Mt. Cargill could explain the persistence of forest right to its summit and also why, where this forest has been recently destroyed, it has been scrub or Juncus caespiticius rather than snow tussock that has taken its place. What the rainfall and evaporation records fail to show is why forest, which can now occur on the lower seaward slopes of Flagstaff and Swampy Hill (where it consists of kaikawaka, mt. totara, and broadleaf) and Maungatua (where it is beech), does not extend to their summits. The sites are drier than Mt. Cargill, but there is no tendency for rainfall to fall further or for evaporation to increase on the snow-tussock slopes above these forests. Indeed the reverse is indicated. A series of stations on Swampy Spur, the lowest just above the forest edge, show this clearly. The following data are for the summer (November 20, 1953 to March 10, 1954) months.

It is difficult to believe that decline in temperature with altitude is proving an effective barrier because of the low altitudes involved and because even in this district the absolute altitude of both kaikawaka and beech is much higher—for kaikawaka, 2,200ft on Mt. Cargill, for silver-beech, 2,400ft on Silver Peak. A sufficient fogtolerance in both species is also demonstrated by these occurrences.

It is clear that the present forest-grassland margins were mainly established over a century ago by a process of deforestation of which log remains and dimples in the grassland provide abundant evidence. It seems likely that the deforestation was aided by a change in climate unfavourable to some tree species, but actually effected by repeated fires of Maori origin. These fires, via an inflammable grassland, always resting against the forest edge, ultimately carried deforestation into parts of this district where forest would otherwise have persisted—though its composition might have changed. Mountain totara, broadleaf, and beech would probably have increased in importance while kaikawaka, pink-pine and mountain toatoa declined. These fires would in general spread from the north-west and have therefore eliminated forest less completely on southern and eastern faces than elsewhere. Holloway accepts the probability of widespread fires (p. 373) and the likelihood of their carrying deforestation into stable forest (p. 378).

However, a possibility that needs to be assessed is that contemporary climate records give no true indication of the climate of these uplands at the time when they became grassland and for most of the ensuing period. Holloway (p. 407) offers some evidence that the South Island climate may recently have become warmer but none that it is wetter. The indifferent regenerative capacity of most existing conifers, and particularly the high proportion of standing dead kaikawaka seem to deny the existence of any recent amelioration of climate, as does the evidence in the

S. Waikouaiti catchment of recent advance of beech into kaikawaka forest. Examination of the longest series of rainfall records available at any single station in this district (Botanic Gardens, since 1913) has revealed no definite trend. There thus seems to be no reason to suspect that present-day measurements are substantially misleading.

The capacity of grassland that is frequently burnt to hold indefinitely areas with a forest climate has been established in other countries (e.g., Beilmann & Brenner, 1951; Richards, 1952, p. 342, p. 350), and if we assume that the grassland has been

Fig. 1.—Improvised rain-gauge.