In certain localities the atmospheric conditions are such that two divisions of the year may be distinguished according to the winds, as for instance in the Marshall Islands, where there are the months of calm and the months of squalls[260]. More commonly two seasons are given by the variation of the monsoons, as on the island of Bali, east of Java: in each case there were six homonymous months. The Kiwai Papuans have _uro_, the comparatively dry season of the south-east monsoon (April-December), and the time of the prevailing north-west wind, _hurama_, a period of alternating calms, storms of wind and rain, and thunder[261]. A native judge from the island of Vuatam in the Bismarck Archipelago remarked that the north-west trade blew throughout the time when the sun was southerly, that is from November to February, but during the time in which the sun moved in a northerly direction, May to August, the south-east monsoon prevailed. On Valam it is said that the south-east monsoon blows as long as the sun sets WNW, i. e. from May to August: from the month of November to February, when it sets WSW, the north-west trade blows[262]. In Rotuma or Granville Island near the equator periods of six months are reckoned. The west wind, which blows from October to April, serves to distinguish these two periods, although it does not affect the vegetation[263]. The people of the Nicobar Islands reckon by the south-west monsoon (November to April)[264]. The Benua-Jahun of the Malay Peninsula distinguish the half-year of the north monsoon and that of the south monsoon[265]. It would seem that the whole year might easily arise through the fusion of these two larger periods: that this is not the case will be shewn in the following chapter. These half-years are as a rule well defined, but the natural conditions upon which they depend are subject to fluctuation, and in particular there are transitional periods the position of which cannot be certainly decided. Moreover smaller characteristic periods arise within the larger, and hence more seasons appear. Elsewhere the natural conditions are such that they directly lead to more than two seasons, e. g. where there are two different rainy seasons in the year. From these circumstances it becomes plain that a fluctuation between a larger or smaller number of seasons is possible, and indeed it often actually occurs. The seasons that adhere to natural phenomena are never clearly defined like a division of the calendar: the limits are uncertain, different seasons may be merged into one another or in part overlap one another, as has been shewn in the case of the Eskimos of Labrador.