Chapter 6.11 Seasonal cycles and lunations

1. Introduction⇫

This chapter complements the ways of talking about time discussed in chapter 9 of volume 2. That chapter addressed concepts like ‘day’ and ‘yesterday’ and ‘tomorrow’. This one is concerned with longer periods of time.

Time is an abstract concept, expressed in western terms by countable named entities, days of the week, months of the year, numbered years. In fact it is difficult for westerners to conceive of time in other than named measurable periods with clear boundaries. But traditionally Oceanic speakers seem not to have thought of time in this way. Here we look at how early Oceanic speakers conceived of years, seasons and the lunar cycle, and consider whether they treated the latter as a system.

The chapter is organised as follows: First, in §11.2, we explore the concepts of years and seasons. In §11.3 the interrelationship between lunar months and the solar year is discussed as their interaction is relevant to any form of calendar. Next, in §11.4, we discuss the checkpoints that recur in Oceanic speakers’ talk about the calendar. In §11.5 we review lunar month names in a range of Oceanic languages, illustrating the use of checkpoints and the kinds of conceptual world they denote. Lunar months imply moon phases, and these are examined in §11.6.

We observed in vol.2(320) that lunar month names “have complex associations with their users’ culture, both material and non-material” and wrote that they would receive a chapter to themselves in a later volume. We endeavour to keep that promise in §11.5 but lunar ‘months’ turn out to be something other than we might have envisioned when volume 2 was compiled.

2. Years and seasons⇫

Early Oceanic communities were inevitably aware of the annual cycle evident in movement of the sun and stars, in regular seasonal changes to weather patterns and in the growth and flowering of plants, both cultivated and wild. They knew that certain foodstuffs, particularly the yams on which many communities were dependent, had a regular growing season, with optimal times for planting and harvesting, and they associated yam planting and harvesting with, among other things, the positions of certain stars in the night sky.

But we cannot be sure that early Oceanic speakers had a concept of, or at least a term for, one complete annual cycle as illustrated by the sun and stars. When we look up ‘year’ in dictionaries we find expressions that translate as ‘rainy season’ (Kove) or ‘yam season’ (Sa’a) or ‘yam harvest’ (Kwaio) or ‘time of ripe canarium almonds’ (Bugotu) or even just ‘garden’ (Gedaged). In other words, examples of well-recognised seasonal cyclic events are used to represent the annual cycle.

Another reason for wondering if speakers had a concept of year was that a person’s age in years was apparently irrelevant to their thinking. To have a meaningful count in years, it is necessary to have a shared base line from which to start counting, something for which there is no evidence across Oceanic communities prior to western influence. Numerous ethnographies refer, both directly and indirectly, to the fact that Oceanic speakers seemed to have no interest in counting years of age (§14.1.2.3). Alkire, for example, notes (1970:37) that “an individual does not think of his age in terms of years (a unit of measurement of little traditional importance in Woleai) … but only comparatively, as being younger or older than some other person of reference.”

Below is the cognate set given in vol.2:308–309 in support of POc *taqun, along with two glosses, the first given in vol. 2, the other a revised definition for which we argue below.

Here we have a reconstruction with sufficient evidence to support three concurrent meanings: ‘year’, ‘season’ and ‘yam season cycle’. But some ethnographic comments give us pause. Codrington writes that in Mota, in the Banks Islands of northern Vanuatu:

There is no native notion of a year as a period of fixed time; the word tau or niulu, which corresponds most nearly to the word ‘year’, signifies a season, and so now the space of time between recurring seasons: thus the yam has its tau, its seasons of five moons from the planting, when the erythrina is in flower, till the harvest, after the palolo has come and gone;¹ the breadfruit has its tau during the winter months; the banana and the cocoanut have no tau, being at all times in fruit.’ (Codrington 1891:349)

Fornander (1878:124) records the following for various parts of Polynesia:

In all the Polynesian dialects the primary and original meaning of tau is ‘a season; a period of time’. In the Tonga group it has the further sense of ‘the produce of a season’ and derivatively, ‘a year’. In the Samoan group, beside the primary sense of ‘season’ it has the definite meaning of ‘a period of six months’, and conventionally that of ‘a year’. In the Society group it simply means ‘a season’. In the Hawaiian group, when not applied to the summer season, it retains the original sense of an indefinite ‘period of time’, ‘a lifetime’, ‘an age’, and is never applied to a year; its duration may be more or less than a year, according to circumstances and the context.’

Their evidence suggests that POc *taqun did not refer to a fixed period of time, but to a period that varied with context. In other words it was a name for any regularly recurring seasonal period. When we find languages using a reflex of *taqun to refer to ‘year’ it seems that, as in the Maori definition, it is being used as just one particular recurrent cycle: its length is irrelevant. A quote from Jenness & Ballantyne (1920:160), writing about the Bwaidoga (PT) speakers of Goodenough Island in the D’Entrecasteaux Archipelago, reflects a similar concept, albeit with a different term. They write that

a native who wished to date some event that happened some time in the past might say that it occurred three malamala ago, in the avalata (north-west monsoon) season, i.e. between October and March;² or in the yam time, from June to August; or he could be still more precise, and name the actual stage in the growth of the yams, and thereby narrow the period down to a single month.

They tell us that malamala is the name given both to a season ushered in by the sun at its northern zenith and also to the whole period covered by the sun’s annual movement, i.e. a year. Chowning’s (n.d.) dictionary of neighbouring Molima confirms the first definition: ‘period from December to April, time of big sun after planting, yamless period’. The ‘year’ definition is confirmed by Huckett, Lucht and Awadoudo’s (1992) dictionary of Iduna, a dialect of Bwaidoga, where malamala is glossed as ‘year’. However, malamala has a deeper history, as its Kilivila and Muyuw cognate is milamala ‘palolo worm’, the risings of which were and are an important checkpoint in the Oceanic year (§11.4.3). At some point in the past, a Bwaidoga speaker who said that something occurred three malamala ago was saying that it occurred three palolo risings ago.

Our conclusion is that POc speakers had a concept of a recurrent seasonal cycle, *taqun, which could be applied to cycles of different durations including possibly the annual cycle. The corollary to this is that there was no POc term for the concept of a fixed, measurable block of time that westerners refer to as ‘year’.

It is difficult to determine which terms for a year in Oceanic dictionaries reflect the longstanding usage of a term and which reflect an adjustment to the western fixed-term concept. Where terms for a year other than reflexes of *taqun have been adopted they typically come from horticulture or arboriculture, such that a salient annual event doubles as the term for an annual cycle. Examples in addition to those given on p297 include the following: In Dobu (PT) yakʷara, ‘last year’s garden’, has become the term for a year. In Kiriwina (PT) the term for the staple crop, taytu ‘small yam’, also carries the meaning ‘year’. Tolo (SES) uvi means both ‘yam’ and ‘year’. In the southeast Solomons languages spoken in Malaita and Makira there is a cognate set³ that refers to the yam harvest but includes ‘year’ among its senses: Lau falisi ‘garden, yam harvest, year’, Kwaio falisi ‘yam harvest, year’, ’Are’are harisi ‘grass, small clover, yam harvest, year’, Arosi harisi ‘year, season, crop’, Sa’a halisi ‘harvest, crop, time of ripening, yam season, year’. The ’Are’are, Arosi and Sa’a dictionaries make the proviso that the denotation ‘year’ is a recent one. Other food crops with regular planting and harvesting seasons also assumed the added sense ‘year’. Thus Marovo (MM, New Georgia) buruburu ‘Canarium spp.’ also means ‘year’, the interval between two ripenings of canarium almonds (Hviding 2005:107). Similarly Maringe finoɣa ‘canarium harvest’, and To’aba’ita ŋali ‘canarium nut tree and fruit’ all also have the sense ‘year’.⁴ In Mangaia in Eastern Polynesia, “in the premissionary times, the age [of a child] was counted by counting the number of breadfruit (kuru) harvests” (Shibata 1999:110).

There are, however, indications that a common application of *taqun was to the complete cycle of yam growing. From the cognate set we learn that the year was equated with the yam season cycle in Buang, Tongan and Maori. Kirch & Green (2001:267) comment that “in Western Polynesian languages, reflexes of PPN *taqu (e.g., [E Futunan] taʔu) refer not just to ‘season’, but more specifically to ‘yam season’.” Although yams, particularly Discorea alata, and taro, Colocasia esculenta, were both important staples for POc speakers (vol.3:256), it was Discorea alata whose time of planting was critical. Taro is not seasonal, growing throughout the year. When people wanted to refer to a time equivalent to ‘last year’ or ‘next year’, they tended to do it by referring to their previous yam garden or their future yam garden.

In the light of this discussion the revised definition of POc *taqun given above reads ‘recurrent seasonal cycle, especially yam season cycle’.

2.1. Named seasons⇫

While POc *taqun was the generic term for any seasonal cycle, specific seasons were separately named. The POc homeland in the Bismarck Archipelago, and indeed much of western Oceania, experience two seasons, strongly marked by wind and weather: the dry, when the southeast trades blow with reasonable consistency, and the wet or monsoon, when the less reliable northwesterlies blow. The names of the winds, POc *raki ‘southeast trades’ and POc *apaRat ‘northwest wind’⁵, almost certainly also denoted the dry and wet seasons respectively. The wet and dry seasons do not have sharp boundaries, however: the focus is on the events that define them.

Almost all Oceanic communities for which we have relevant information divide the year into these two main seasons, sometimes accompanied by a short season between them, but the names do not always reflect the POc terms. Thus Wogeo had the kama ‘trade wind season’ and the yavara, ‘monsoon season’ (< POc *apaRat) (§11.5.1.1). Maenge had vinte ‘the wet’ (May–September) and kaepâ ‘the dry’ (November–March) (§11.5.1.2). Barok has awat (< POc *apaRat) “identified with the traditional cycle of six lunations, [while] two awat are encompassed by the sun’s annual circuit of the ecliptic” (Wagner 1986:40). The two awat are awat ni nien ‘season of plenty’ and awat nere loŋ ‘season of hunger’. Barok awat has thus taken over the semantics of POc *taqun. Tangga, spoken on small islands east of New Ireland, contributes no month names but offers bāt ae us, ‘the rainy season’ (November–March; us ‘rain’) and pisae ‘the sun’ or ‘dry season’ (May–September) (Bell 1946:143).

Some of these names may have originally been allusions to crop-based seasons. Lichtenberk (2008:177) writes that the To’aba’ita year was traditionally divided into two halves: six months of canarium almonds (April–September) and six months of strong winds (October–March). Ivens (1927:397) writes that in Sa’a “practically there are two divisions of the year: marāu or āu, the time associated with the canarium almonds [ŋali], and oku, the time associated with the palolo” (cf §11.5.4). Mota had maɣoto ‘Miscanthus grass’ (‘wet season’), rara ‘Erythrina’ (‘dry season’), and ud ‘palolo season’. Kirch & Green’s (2001:260ff) detailed examination of time reckoning and the ritual cycle in Polynesia is summarised in 11.5.7.

Grimble (1931) describes a Kiribati year of two seasons, marked by observation of te auti ‘Pleiades’ from early December to early June,⁶ and rimʷimāta ‘Antares’ from early June to early December. The year is considered to begin with the appearance of the Pleiades about 15 degrees above the eastern horizon just after sunset, in about the first week of December. It seems likely that Kiribati time reckoning has been influenced by Polynesian: two seasons are maintained, but are now star-based.

A number of names like those above also serve as labels for what ethnographers sometimes describe as ‘months’, but we argue in §11.5 that these ‘months’ do not add up to anything like a calendar in the modern Western sense.

3. Lunar months and solar years⇫

There is a tension between the solar year with its solstices and the lunar cycle. The dry and wet seasons and the times for planting and harvesting crops are all governed by the solar year and the stars, while shorter periods of the year are linked to named lunar ‘months’. How was this difference reconciled in traditional Oceanic societies?

A lunar cycle consists of 29.53 days. Twelve lunar months total a year of 354 days, eleven days short of a solar year. In most Oceanic societies for which we have descriptions of lunar months, the year was divided into twelve lunations, although in some cases thirteen are recorded, totalling 384 days. Either way, if a strict program of activities were carried out according to the lunar calendar it would have gradually become out-of-sync with the annual seasonal cycle. It follows that a system of reckoning time by referring to a systematic list of lunar months has little practical value unless it includes some mechanism for intercalation, i.e. for inserting days to align the lunar year with the solar year.

The need for intercalation carries with it an assumption that people are aware of the existence of a solar year containing a specific number of lunar months. But this assumption is not borne out by the few detailed accounts we have of attitudes to time, at least in WOc. While most ethnographers evidently assume a concept of a year as a fixed period of time, for speakers time is more flexible, with allowance for adjustment so that lunar months and known seasonal events do not get out of step. Malinowski reported (1927:209) that “in the Trobriands the moons are used rarely and only under special circumstances for counting time; the whole system of naming and arranging moons has no special place in their time-reckoning.” Rather they would become aware that at times the moons and the start of a seasonal cycle were out of step, or as the Trobrianders put it, “the moons become silly” (1927:213).

Damon (1990:35) distinguishes between lunar month names (kʷel) and moons in Muyuw (PT).

Muyuw can count moons. They do not count kwel (although a new kwel begins with a new moon)… People might be able to tell me at what kwel something should happen but not how many kwel between now and then. They are keen observers of the moon and its phases, but they do not systematise their observations.

Damon (1990:4) considered the question of intercalation, noting that while the east had a twelve month calendar, Central Muyuw had a thirteenth month. He writes “This disparity might once have meant something quite significant – perhaps a way of adjusting solar/lunar discrepancies … Repeated attempts to explore this contrast, however, revealed nothing during either of my research periods.”

Chowning & Goodenough (2016) writing of the Nakanai of New Britain who identify eight months and Seeman (1862:297) who had considered the problem inherent in his eleven-month Fijian calendar, both describe similar solutions. In both locations a period of from two to four months is treated flexibly so that the rest of the year is marked out by more precise markers of time such as the palolo rising (§11.4.3). In Nakanai this period is about four months, roughly from June to September. In Fiji the period is aligned with the time of clearing and preparing gardens around June and July.

Wagner (1986) writes that the Barok of New Ireland use correlation between the moon, the sun and the Pleiades in regulating their gardening activities, with the movements of the relatively constant sun and Pleiades acting to correct the seemingly variable nature of the lunar cycle. He names six lunations, suggesting that one, tege gowo, ‘the one that is left’, may serve to fill the variable space until the appropriate new moon appears (§11.5.3.2).

In Polynesia, where the calendar became more systematised than in early Oceanic, there were procedures for intercalation. According to Collocott (1922:168) in Tonga

With [the month of] Tanumanga the year normally ends. If, however, observation of the yam and other plants and of fishes at the next new moon fails to discover the appearances proper to the month with which the year begins, Lihamua, another month is intercalated.

Buck (1932:230) records a different system in Rakahanga in the Cook Islands whereby the usual year of twelve lunar months would be replaced at intervals by a thirteen-month year. “The intercalation of a thirteenth month was decided by the simple rule that a new year could not start until the first new moon after the morning rising of the Pleiades”. The strict application of the rule would automatically lead to the intercalation of a 13th month in some cycles.

In Hawaii according to Makemson (1941:97) a calendar consisted of twelve months of 30 days plus five days interspersed at various times set aside for religious rites. We wonder if this is in fact a post-contact adjustment. We have found no evidence elsewhere in Oceania that the year is seen as a unit of fixed length of 365 days, but rather a collection of recurring cycles.

The assortment of ways in which communities, or perhaps ethnographers, have tried to fit lunar months into a solar year leads one to think that not only was this difficulty unanswered in POc times, but rather that the problem simply did not exist for its speakers. If the annual cycle was seen as not a fixed period of time (§11.2), and if lunar months could not be combined to form an assembly (§11.5), the question of reconciling the two systems becomes meaningless. Rather, the topic may have become a matter for debate following introduction of the western conceptual system of time.

4. Checkpoints⇫

The term “checkpoint” is used here of natural, arboricultural and horticultural phenomena that occur cyclically and allow Oceanic speakers to locate themselves in the cycle of the seasons and prompt them to perform particular activities. Some checkpoints are fairly precise; for example, the rising and setting of the Plaeiades (§11.4.2.1), the annual rising of the palolo sea worm (§11.4.3). Other checkpoints are fuzzy: for example, the readiness of the canarium almonds for harvesting (§11.4.5), or the beginning and ending of the wet season (§11.4.7).

As becomes obvious below, a checkpoint can also sometimes become entrenched as the name for a period of time in which the named phenomenon occurs. When we use the term ‘month’ in an Oceanic context, it is these periods of time that we are referring to: not a period of time defined by its boundaries but a period of time that centres on a cyclically occurring event (see further §11.5). Exceptions to this generalisation are found in Micronesia (§11.5.7) and Polynesia (§11.5.8).

Most of our data on seasonal time are in the form of lists of so-called lunar month names collected from more than 30 languages from Western Oceanic, the SE Solomons, Vanuatu, Micronesia, Fiji and Polynesia. They consist of, usually, 12 or 13 names for or references to the kinds of phenomenon mentioned above that serve to mark roughly sequential points or approximate periods of time through an annual cycle. The lists show that the same kinds of markers are recognised in widely scattered parts of Oceania, notwithstanding its geographic range. They include the apparent annual movement of the sun (§11.4.1) and stars (§11.4.2), the palolo rising (§11.4.3), plant cycles (§11.4.4–6), weather patterns (§11.4.7), and land crab migrations (§11.4.8). In spite of this, the lists are a very varied lot. It is apparent that except in Micronesia and Polynesia they offer very little in the way of shared terms that could be taken as a basis for reconstructing a calendar of POc lunar months. In this chapter we explore the ways in which communities used these checkpoints in order to see if any system can be recognised.

4.1. Solstices⇫

Solstices occur twice each year, in June and December. In the Bismarck Archipelago and elsewhere in the southern tropics, the June solstice is the time when the sun reaches its lowest zenith, the December solstice its highest. Their occurrence as checkpoints in Oceanic communities is mentioned infrequently in the literature, but all early Oceanic speaking communities must have been aware of them.

Panoff (1969:155) writes of speakers of Maenge (NNG; a dialect of Mengen, SE New Britain):

Both extreme points from which the sun rises at solstices are perfectly known and are identified with conspicuous landmarks on the horizon (mountain, reef, islet etc.), which differ from village to village according to the surrounding topography. They are called kae taraŋana ‘resting places of the sun’, since they correspond to a ten days’ full stop in the shift.

Jenness & Ballantyne (1920:160) write that the Bwaidoga (PT)

have noticed [the sun’s] annual movement and related it to their gardening operations. Igoboda, the time when the sun is farthest south, is the period when gardening commences; when it reaches half-way back to the north again it is harvest-time; and at its northern zenith it ushers in the malamala season.

Wagner (1986:39) writes that the Barok (MM, New Ireland)

seem always to have noted the seasonal variation in the points-of-rising of the sun and moon … The northern solstice occurs when the sun rises over Lihir and the full moon rises over Namarodu … and the southern solstice when the sun rises over Namarodu and the full moon rises over Lihir.

Although similar terms for the solstices have been noted in various parts of Polynesia (see vol.2:153), no reconstructions are possible.

4.2. The stars⇫

Stars appear to move across the sky in a circle whose centre is the north or south celestial pole. Stars closer to the pole describe a smaller circle and never disappear from the night sky. Others describe a larger circle that takes them below the nighttime horizon: they traverse the sky invisibly during sunlight. The first pre-dawn rising of a star after a period of invisibility and the last post-dusk setting before invisibility had calendrical significance for many premodern peoples.⁷ But the event recognised in pre-modern communities is the first brief apparent pre-dawn rising of the star, which occurs when the star is high enough above the horizon to be seen, perhaps two weeks later than its astronomical counterpart. Apparent risings are later than their astronomical counterparts and apparent settings earlier. On the basis of dates given in ethnographies, we assume that the apparent first or last rising/setting occurs when the star is 15 degrees above the horizon, but the actual date depends on the topography of the community’s environment and on weather.

These events affect all stars that rise and set. They appear to have been of especial importance in Nuclear Polynesian (§11.5.8), where certain month names are associated with the rising or setting of a given star. PNPn *tolu and PEPn *takulua evidently referred respectively to the pre-dawn rising of the middle star in Orion’s Belt and of Sirius, PNPn *tākelo to the post-dusk rising of Betelgeuse. The most important sidereal checkpoints for traditional Oceanic speakers were the risings and settings of the Pleiades (§11.2.1), but it can scarcely be the case that the Pleiades were the only night sky events of which they took notice.

With the single exception of the Pleiades, knowledge of stars and their movements seems today scarcely to exist in western Oceanic communities.⁸ The only record there of stars being used as calendar reference points comes from two closely related communities, Kilivila speakers of the Trobriand Islands and Muyuw speakers from nearby Woodlark Island (§11.5.2.1.1). A far more detailed awareness of stars has been retained in Micronesia and Polynesia, no doubt due to their importance in navigation (Lewis 1972) (vol.2, ch.6). Here one finds month names that are simultaneously star names. These terms, however, are never cognate with those in Kilivila and Muyuw.

4.2.1. The Pleiades⇫

The Pleiades are a small bright patch of stars with an annual orbit such that at times they disappear from the night sky. Their significance as a checkpoint may have formed part of the corporate memory that the ancestors of POc speakers brought with them from a former homeland. Their presence with similar functions in languages across the Indo-Malaysian archipelago (Forth 1983; Ammarell 1988) suggests that this was true at least as far back as PCEMP.⁹ The Pleiades have been recognised as significant calendrical markers throughout the Oceanic world, although the timing of their appearance, and thus the particular event(s) they mark, have changed by about five weeks in the approximately 3000 years since POc was spoken¹⁰.

The dates of the Pleiades’ apparent risings and settings in the year of writing (2016) and in 1200 BC (approximately when POc was spoken) at Kimbe, New Britain, are given in Table 11.1. Kimbe is chosen as it is within the assumed region occupied by POc speakers. In Apia, Samoa, the dates are just a day later.¹¹

Areas in which the Pleiades are known to provide checkpoints include the north New Guinea coast (Wogeo, Table 11.10 in §11.5.1.1), around the Huon Peninsula of New Guinea (see below), in Mangap-Mbula (Table 11.12 in §11.5.1.3), in Barok (Table 11.18 in §11.5.3.2), in Micronesia (Table 11.22 in §11.5.7) and Polynesia (Table 11.23 in §11.5.8.1).

In Western Oceanic communities the position of the Pleiades in the night sky provides a series of indicators to stages in the yam cycle. In a number of languages of the north coast of New Guinea, the Pleiades (Gedaged bal̥ as, Bing barahas, Takia baras, Wogeo baras, Manam barasi) are associated with young women and fertility rituals marking the start of the agricultural cycle. It is worth quoting Mager’s (1952:17–18) Gedaged dictionary entry for bazas (= bal̥ as) in full. It refers to the Pleiades’ first pre-dawn rising.

The Pleiades constellation thought of as young unmarried women. When they reappeared on June 13th or 14th the fertility rites were observed. When first seen the tauz triton shell was blown and a big rumpus made by beating and shouting. All the young people were awakened and driven into the sea to bathe; this was to cause them to be healthy, tall and beautiful. … When the Pleiades reappeared the people knew that it was time to prepare the fields for planting yams.

Hogbin (1938b) describes the situation in Wogeo (§11.5.1.1), an island off the north New Guinea coast, where certain rites known as baras losalosa ‘washing the Pleiades’ (or ‘washing the pubescent girls’), are performed.

These are associated with the changing position of the Pleiades, a constellation known as Baras, the term for a girl passing through her first menstruation. … The purpose of the rite is to secure protection from sickness during the coming year and to ensure a good nut harvest (1938:138).

Wedgwood (1934-35:397) describes an apparently identical rite in nearby Manam, barasi di-ruʔu ‘they wash the Pleiades’, marking the beginning of the agricultural year. It takes place in the months of April, May or June according to the village which is performing it. Wedgwood comments

I was not able to find out how the people of Manam adjust the lunar year to the solar year, but I was given the names of thirteen ‘moons’, and was told that the people knew which ‘moon’ was which by the position of the Pleiades just after sundown in relation to the mountain top (Wedgwood 1934:397).

Wedgwood’s ‘mountain top’ refers to Manam Island itself, a near-perfect volcanic cone.

There is a tantalising entry in Chowning’s (n.d.) dictionary of Molima (PT): veʔovaiya-liwoliwo ‘to greet the reappearance of the Pleiades’ (presumably its first pre-dawn rising), but apart from ovaiya meyavinena ‘Pleiades’ there is no elaboration. The word ovaiya has no known meaning outside this context, but meyavine-na means ‘female’. Is this an echo of the association between the Pleiades and pubescent girls attested in Gedaged, Wogeo and Manam? Intriguing is the fact that ovaiya meyavinena ‘Pleiades’ contrasts with ovaiya meʔolotona ‘Orion’s Belt’, where meʔolotona means ‘male’.

In Yabem (NNG) Streicher (1982:80) writes about the period when the Pleiades are prominent in the night sky:

the Pleiades [dam, damɔ] are the main constellation seen by the Jabêm [Yabem] during the dry season (October to March) and governing their activities in their gardens; i.e. the felling of trees to clear the ground for new gardens; the burning and planting of fields is done according to the position of the Pleiades.

Along with other terms for the Pleiades from around the Huon Peninsula, Yabem dam, damɔ appears to reflect PAn *damaR. However, PAn *damaR is also more obviously the ancestor of POc *ramaR ‘coconut leaf used as a torch when fishing’ (vol.3:382). The Huon Peninsula terms would reflect a putative POc *dramaR, of which the initial nasal-grade consonant would probably reflect an unknown morphological modification. Its POc reconstruction remains uncertain.

Yabem dam-saŋiŋ is a compound which Streicher (1982) explains as follows:

People ask each other, “Have the Pleiades disappeared from the western sky or not?” Disappearance of the Pleiades marks the end of the dry and the beginning of the wet season. Hence dam ‘Pleiades’ and saŋiŋ ‘enquiry’.

The Yabem term thus refers to the last post-dusk setting of the Pleiades in late April or early May, and the Numbami term evidently marked the same seasonal transition. The Sio and Mangap terms, however, seem to refer to their first post-dusk rising in early December, and an ethnographic note in Bugenhagen & Bugenhagen (2007b) says that this marked the canarium trees beginning to form buds (Table 11.12).

Although the Pleiades carry less weight for the Maenge of New Britain (NNG), their appearance and disappearance are noted. Panoff (1969:156) writes that

The movement of the Pleiades, which are called kumana puni me, literally ‘a dense cluster of young taros’ has failed to suggest to the Maenge the notion of a yearly cycle, although their disappearance [their last post-dusk setting–MO & MR] is interpreted as a signal to plant the last taros before the heaviest rains of the wet season.

In Bwaidoga (PT) in the northern D’Entrecasteaux, Jenness & Ballantyne (1920:161) write that the Pleiades

is the best known of all the constellations…. The natives often date their yam harvest from the time when the Pleiades appear in the east in the early evening till the time when they have moved over to the west.

In Dobu (PT), Fortune (1963:127) writes that

gathering times are regulated by the position of the Pleiades in the sky … When he rises at about 15º angle with the ocean the bush is cleared; at about 30º the land is planted. He climbs from the north-eastern to the south-western sky, sets in the south-west, and is unseen for over a month. Then, when he rises in the north-east, harvest time is come.

Damon (1990:39) notes with regard to Muyuw speakers (PT; §11.5.2.1.2) that yams should be planted when the Pleiades “is thirty or so degrees above the western horizon at dusk, in February” en route to their last post-dusk setting.

For the Barok people of New Ireland (MM), “the timing of the gardens is regulated by three celestial indicators: the moon, the sun and the Pleiades” (Wagner 1986:37). A discussion of the Barok calendar is provided in §11.5.3.2.

In his discussion of local knowledge of the heavenly bodies in north Vanuatu, Codrington (1891:348) writes that

The Banks’ Islanders and Northern New Hebrides people content themselves with distinguishing the Pleiades, by which the approach of yam harvest is marked.

The Pleiades play a significant calendrical role throughout Micronesia where it takes its place in a sequence of twelve stars or constellations that serve as monthly timekeepers (§11.5.7).

The calendrical uses of the Pleiades described above refer to single events. In some EOc languages, however, this has developed into a marking of the two seasons into which the year is divided (§11.3).¹²

Kirch & Green (2001:260ff) have made a detailed examination of the reckoning of time and the ritual cycle in Polynesia, and we have drawn on their account for much of the following. They quote from early descriptions – Tahitian King Pomare in 1818 (quoted by Henry 1928), and Gill (1876) on Mangaia, among others – showing that the risings and settings of the Pleiades were widely observed in many Polynesian societies, “where they were used to mark the change in seasons and/or to mark the commencement of the year” (Kirch & Green 2001:262). In this they concur with Makemson (1941:76) who wrote that

undoubtedly the Polynesians carried the Pleiades year with them into the Pacific from the ancient homeland of Asia. With but few exceptions they continued to date the annual cycle from the rising of these stars until modern times.

Gill (1876:317) writes that in Mangaia

The arrival of the new year was indicated by the appearance of Matariki, or Pleiades, on the eastern horizon just after sunset, i.e. about the middle of December. Hence the idolatrous worship paid to this beautiful cluster of stars in many of the South Sea Islands. … In many islands extravagant joy is still manifested at the rising of this constellation out of the ocean.

Kirch and Green note (Kirch & Green 2001:261–262) that the pre-dawn rising of the Pleiades is observed as a significant event in East Futuna, Tikopia, Rakahanga, Pukapuka, Mangareva, Tuamotu and New Zealand, while the post-dusk rising is significant in Tokelau, Tuvalu, Tahiti and Mangaia. Both are significant in Hawaii. The pre-dawn rising counts as the beginning of the year in Tokelau and Rakahanga, while in other locations across Polynesia it is the post-dusk rising that counts. Kirch and Green infer from these distributions that both dates were important for the early Polynesians, the May and October risings – or the new moons which followed them – marking the beginnings of the two seasons, PPn *taqu, into which most Polynesian communities divided their annual cycle. Strictly, it was probably not the rise of the Pleiades themselves that counted as the beginning of the new season, but the appearance after their rise of the first sliver of a new moon.

As Kirch and Green note, the ancestral Polynesian calendar was inseparably linked to the horticultural year, and especially the seasonal yam crop, whose scheduling depended on climatic seasonality within the Polynesian homeland (Kirch & Green 2001:265). The pre-dawn rising in May signalled the onset of the dry season in the Tonga-Samoa region, the Polynesian homeland, while the post-dusk rising in October announced the onset of the wet.

A POc term for the Pleiades, repeated here from vol.2:171, can be tentatively reconstructed – “tentatively” because the Nakanai, Roviana and Gela reflexes are phonologically irregular. However, no reflexes have been found in calendrical terms.

Apart from the small groups of New Guinea north coast and Huon Peninsula languages mentioned above, terms for the Pleiades in Western Oceanic show no evidence of cognacy. They are reconstructable as month names only in Micronesia (PChk *mʷakariker ‘about July; the Pleiades’; §11.5.7) and Polynesia (PNPn *mataliki ‘month name, June’, from PPn *mataliki ‘Pleiades’; (§11.5.8).

4.3. Palolo risings⇫

The palolo worm, Eunice viridis (also Leodis viridis or Palolo viridis) is a segmented sea worm that lives in crevices in a coral reef. Its annual spawning occurs in a widely distributed number of places, but always at a time associated with the lunar cycle. In Eastern Oceanic communities this is typically six to nine days after the full moon in October and November (Burrows 1955:141) while in Wogeo Hogbin (1938b:132) identifies it with the week preceding the full moon. There is a minor rising, followed one lunation later by a major rising. Occasionally, if the first rising occurs very late in the cycle there may be no second rising. The lunar cycle moves back about eleven days every year, so that the critical dates actually occur, as far as there are records, at variable dates between mid-October and mid-December (Burrows 1955; Caspers 1984).

Each year, before spawning, the palolo generates a tail, often several times larger than its body, containing eggs and sperm. When the lunar timing is right the tails are released and undulate to the surface where they form a writhing mass, the segments bursting and releasing a milky, gelatinous soup of eggs and sperm. A rising lasts about four hours before dissolving away. Its occurrence is predictable, and in many Oceanic locations the palolo’s breeding frenzy is interrupted by people who scoop up the tails before they burst and cook them as a culinary delight.

The palolo’s appearance, although recorded in various places in Western Oceanic, carries less calendrical weight there than in Eastern Oceanic. Damon reports that on Muyuw (Woodlark Island) although the palolo (milamala) appears around the full moon of, usually, October, it is neither eaten nor used for any calendrical purposes (1982:229). Similarly, although Maenge on the south coast of New Britain

seem to be able to foretell with good accuracy the time of appearance of Palolo viridis, they have never thought of making a time marker of this striking phenomenon (Panoff 1969:158–59).

However, Hogbin (1938b:132) writes for Wogeo on the NNG coast that

the seven or eight days preceding the night of the full moon in late October or early November are deliberately avoided when fixing dates for festivals. On this one night of the year (or occasionally on the night following) that curious marine annelid, the palolo worm [manuam], rises to the surface of the sea for spawning. It is regarded–with reason–as a great delicacy, but the haul is so uncertain that a taboo is imposed beforehand on all save urgent tasks in an attempt to ensure the co-operation of supernatural forces in securing favourable conditions.

Mondragón (2004:294) writes that in Loh in the Torres Islands of northern Vanuatu the palolo rising is not simply a source of food but has elicited long-standing ritual.

Once a year at dawn (on November 15th by the Gregorian calendar), if the sky is clear, the people of Loh gather in Peliauluwo to observe the rising sun as it emerges just to the south of the outline of Ureparapara. This, they claim, is a signal that the Palolo shall emerge from the ocean later that day. Although there has never been a tradition of horizon-based astronomical observation in the Torres, Loh islanders have long employed the peculiar solar alignment as a key indicator of the emergence of the Palolo and the approach of the summer solstice.

Although there are obvious errors in linking the rising of the palolo with a fixed solar event rather than a more mobile lunar one, as Mondragón notes, the association is presumably a remnant of some earlier calendrical ritual which has now been mistakenly fixed into the western calendar.

Because the palolo worm is tied directly to the lunar cycle, it plays a substantial role in the naming of lunar months. Many languages with month names that include a palolo term have a pair of such names associated with the small and big risings and denoting successive months around October and November. A number (Sa’a, ’Are’are, Loh, Mota, Mwotlap) use the term for palolo to refer to a season that may extend for several more months, while in Kwaio its regular appearance marks the span of a year. The palolo occurs in month names in a few Western Oceanic languages (Kairiru munuan,¹³ Wogeo manuan, Yabem igeyaŋ, Bing yagyahag, Kilivila/Muyuw milamala), but the only reconstructed terms are PEOc *(o,u)du ‘palolo worm’ and PCP *balolo ‘palolo worm, season name’ (vol.4:212). Below is a list of SES and NCV palolo month names. From these it is clear that PEOc would have had a term meaning literally ‘big palolo’ referring to the month of the major rising, with a range of possible terms for ‘big’.¹⁴

Semantically the Fijian languages agree with the languages above in reflecting PEOc ‘big palolo’. They also agree with Loh, Mota and Mwotlap in having month names meaning ‘small palolo’, suggesting that the latter occurred in PROc.

Like Fijian, Polynesian languages reflect PCP *balolo ‘palolo worm’, but differ in replacing ‘small’ and ‘big’ with reflexes of PPn *muqa ‘first’ and *muli ‘last’. Palolo risings have not been reported in Polynesia outside Tonga and Samoa, and month names have become divorced from their original reference and are now simply recurring names in a list. Even in Samoan the terms refer inexplicably to two non-palolo months.¹⁵

What the listings above do not show is that the palolo month names were integrated into a system that included other aspects of the annual cycle, aspects which are sometimes mentioned in the notes attached to lists of names. Thus in Wogeo manuan also marks the beginning of yavara ‘north-west monsoon season’ (< POc *apaRat, vol.2:129–130) while in Sa’a oku peine marks the awalosi (vol.2:130), again the northwest monsoon that brings the rainy season (Hogbin 1938b:137; Ivens 1927:397). Kilivila milamala marks the beginning of the new year, preceded by the yam harvest and followed by the burning, clearing, and planting of next year’s gardens (Damon 1982:231). Similarly in Loh, yams and other crops are planted around nʉt ‘palolo’ and poles are erected for the climbing vines around nʉt lavü ‘big palolo’. The series of nʉt month names continues with nʉt melüɣehe ‘green palolo’ and nʉt memüdarü ‘red palolo’. According to Durrad (1939) and Alexandre François (fieldnotes) ‘green palolo’ is when the yam vines are in full green leaf and ‘red palolo’ is when they begin to turn red. Codrington (1891) records ‘small palolo’ as the season of maturity and ‘big palolo’ as the time for digging up the tubers. Thus his yam calendar runs a month behind those for Kilivila and Loh.

Although no POc term for palolo can be reconstructed (vol.4:212), there is enough evidence to suggest that POc speakers may have had at least a calendrical checkpoint named for the palolo.

4.4. Yam or taro cultivation⇫

The one marker of the time of year that would have been consistently recognised by all members of a community was the stage of development of the staple crop, particularly the yam, Dioscorea spp., on which many communities depended for their survival (vol.3:258–261). Yams are seasonal, with planning essential for prior garden preparation and sowing. Once established, the developing vines are usually given a stake or frame for support. Leaves change colour and die off, indicating the appropriate time for harvesting. The yam crop effectively serves as a kind of time line along which various points may be identified to mark a particular occasion or event. Thus, almost the whole life of a community, the times for greatest gardening activity, for rituals, for feasting, for trading, would be dependent on the timing of the staple crop. So it is not surprising that month names based on the various stages of yam cultivation, are common in Oceanic languages such as Kwaio (Table 11.3), Mota (Table 11.4), Fijian (Table 11.5) and Tongan (Table 11.2), although they do not extend beyond Tonga, which lies at the eastern extreme of cultures practising yam cultivation. Tongan itself, however, has one of the most articulated (or perhaps best described, by Collocott 1922) sets of agriculturally based month names.

At the opposite extreme, there are almost no Western Oceanic month names that directly denote stages of yam cultivation. However, certain Sinaugoro names refer indirectly to yam growing (Table 11.16 in §11.5.2.2). The term ɣʷa-koli ‘April approx.’ means ‘month of no food’ (ɣue ‘moon, month’, koli ‘finished’), referring to the time of hunger, while ɣʷa-ɣaniɣani ‘May approx.’ means ‘month of eating’ (ɣaniɣani ‘eating, food’) heralding harvest after ɣʷa-koli.

Other month name sets that refer to agriculture are from the SE Solomons and Vanuatu, and refer to stages of yam cultivation. Table 11.3 lists Kwaio (SES) month names that refer to such stages.

Loh (NCV) month names are set out in Table 11.20 in §11.5.5. Here the colour terms in n’ʉt mələɣɛhə ‘green palolo (January)’ and n’ʉt meməʈarə ‘red palolo (February)’ refer to the leaves of the yam, not to the palolo.

Codrington (1891) lists a set of alternative month names in Mota (Table 11.4) which clearly refer to gardening activities, and these are echoed in the limited data available from central and southern Vanuatu. Thus Atchin (Malakula) hil-hilɛn ‘name of a feast’, literally ‘digging up’ and ruwan ‘clearing (forest)’. Two other Atchin month names echo the Sinaugoro pair above: boŋ hoal wele ‘days of little food’ and boŋ hoal lɛp ‘days of much food’. Crowley (1998:154) notes that in Sye (SV) some month names refer to the yam cycle. They consist of mov and a verb. Thus mov-ɣorovoh (August) where ɣorovoh means ‘s/he cleared a garden site’ is ‘month of clearing a garden site’. Similarly mov-ɣerevei (September) ‘month of trimming’, mov-ɣowi (November) ‘month of planting’.

There are terms in Wayan Fijian (Table 11.5) that look suspiciously like month names as they begin with vula ‘moon, month’, but it is clear from the period lengths they denote (in the leftmost column) that they here mean ‘season, time of year’, as Pawley & Sayaba (2022) note. The same is probably true of the similar expressions in Bauan Fijian in Capell (1941).

4.5. Canarium arboriculture⇫

Canarium arboriculture was practised in New Guinea probably for millennia before the arrival of Oceanic speakers. There are numerous tree species of the genus Canarium. Those grown by Oceanic speakers are discussed and terms for them reconstructed in vol.3. POc *[ka]ŋaRi referred to C. Indicum and probably was also used as the generic term (vol.3:312–317). The edible fruit is often called the canarium nut or canarium almond.¹⁶

Canarium almonds continue to be highly valued in Near Oceania and Vanuatu because they are good eaten raw or smoked, and pounded almonds are an essential ingredient in much appreciated oily puddings. As a result, their annual harvest in July and August has given rise to various ceremonies and rituals (vol.3:314), and these have come to mark certain months of the year in various communities in Near Oceania.¹⁷

The most extreme instance of canarium-based month names in the data is in Mangap (NNG) from Bugenhagen & Bugenhagen 2007b, listed in Table 11.12 in §11.5.1.3. The ethnographic definitions shown in the “Accompanying notes” column of Table 11.12 all refer to canarium (Tok Pisin galip) trees.

The Roviana month names in 11.4.5 are from Waterhouse’s (1949) dictionary. Only muzara ‘approx. October’ is assigned to a western month. The month haele can be assigned to July–August as its gloss (‘climb’) and the accomanying note indicate that it is the time of the canarium harvest. The month lomu kubata must also be somewhere around harvest time.

Waterhouse’s definition of okete ‘Canarium’ reads

okete, a tree [Canarium sp.]. The ripening, gathering and storing of the nuts were important features in Roviana life, and several of the months take their names from various phases of the okete cult.

Not much appears to be known about the Roviana canarium cult, but the month names in 11.4.5 indicate that canarium was sacrificed at hope, ‘the general name for sacred places, especially where skulls are placed’ (Waterhouse 1949). There is plentiful archaeological evidence of canarium in Roviana shrine excavations (Aswani & Sheppard 2003). A perhaps similar cult is recorded on neighbouring Choiseul by McClatchey et al. (2006a). About canarium, they write:

Its usage so permeated peoples’ ceremonial and dietary lives that some ancient and special relationship is connotated particularly through usage as a sacrificial offering and as a principal symbol of land tenure and authority … The uses and interactions with Canarium by the Babatana and Ririo are extensive and permeate most aspects of traditional life. It is difficult to imagine these cultures in the absence of Canarium.

Various month names referring to canarium almonds in other Oceanic languages are listed in Table 11.7. These mostly denote the time when the canarium almonds are first ready for harvesting, and the geographic distribution of these names further attests to the cultural importance of canarium.

No reconstructions of month names involving the canarium almond can be made, even at quite local levels, but the data and discussion in this subsection indicate that in cultures of Near Oceania the canarium almond is more important than any other food stuffs except the yam (§11.4.4), and that it has (or had, traditionally) a significance that goes beyond nutrition. In the western Solomons there was a canarium cult. Whether or not the naming of months for events in the canarium cycle reflects an earlier canarium cult, is unknown. However, the cultural significance of the canarium is widely enough spread in Near Oceania to suggest that if there was a term in POc-speaking communities labelling a lunar month around June, July and August it was likely to be one that referred to the ripening or harvesting of canarium almonds.

4.6. Wild plants⇫

The Maenge calendar (Table 11.11, §11.5.1.2) represents the most extreme instance of months being named for wild plants, but such names are fairly common across Oceania. One of the most widespread is the flowering of the coral tree or erythrina (POc *rarap ‘Erythrina spp.’, POc *baR[baR] ‘Erythrina variegata’; vol.3:158–161), with distinctive bright orange-red spiral flowers at the end of each branch. It occurs in month names in Maenge (NNG), Sinaugoro (PT; Table 11.16) and Mota (NCV; Table 11.20). The tree is often a salient feature in and around Melanesian coastal villages, and its flowering is often taken as an indicator that it is time to plant yams. Other references to the erythrina are less specific (or less well understood). Two NCV languages of Malakula refer to the erythrina. They are Avava, which has the name ivlemial for November, glossed ‘red, Erythrina variegata’ by Crowley (2005), and Atchin, for which Capell & Layard (1980) give the month names rere tsar ‘the leaves of the erythrina are falling’ (tsar ‘erythrina’) and ni-rere ‘[the erythrina is] red’.

Also featuring in month names are two tall grass species. In Western Oceanic this is usually Saccharum edule (PWOc *tabuqaR, vol.3:301), known in Papua New Guinea as pitpit, a tall grass related to sugarcane (Saccharum officinarum). It is cultivated for the unopened inflorescence at the tip of the cane, which is harvested as a seasonal vegetable that is either roasted in its leafy sheath or cooked in coconut cream with other vegetables. The other grass is Miscanthus floridulus (POc *pi(y)uŋ, vol.3:253), a reed-like grass which grows on dry hillsides, to about 2 metres tall.

Table 11.8 shows that the month names that refer to the two grasses are more or less in complementary distribution, with Saccharum edule in NNG and Miscanthus floridulus in SES, Vanuatu and Fiji. The event that is marked across the languages in Table 11.8 is the flowering of the grass, usually in April or May (Bariai, Yabem, Arosi and Wayan Fijian). It is at least plausible to suggest that this month was also labelled in this way in POc, and that terms for Saccharum edule were replaced by terms for Miscanthus floridulus as Oceanic speakers moved eastward.

4.7. Weather patterns⇫

The weather cycle of Pacific communities is dominated by the contrast between the wet and the dry seasons. In the dry season, typically from May to October, the southeast trades (POc *raki) blow, in the wet the northwesterlies (POc *apaRat), from December to March, and the sea is rough (vol.2:131–135). Between the seasons are the doldrums.

Month names in some languages make fairly simple reference to the seasons. Thus in Kove (NNG) May is called hai kaŋkaŋa, presumably the start of hai (< POc *raki), August simply hai, and the period from September to November sua-hai (Chowning 2009). Sinaugoro and Motu (PT) each have a month lailai, but in Sinaugoro it denotes June, in Motu March. Sinaugoro also has the term lai-toɣa ‘quiet lai’ for the June–July period.

’Are’are (SES) has several month names that refer to the dry, calm season ōku: ōku rate ‘Miscanthus dry season, September’, ōku māʔa ‘very dry season, October’, ōku paina ‘big dry season, December’ (Geertz 1970).

Fewer languages have month names reflecting POc *apaRat. Those that do are on the south coast of Papua: Lala avala ‘thirteenth month’ (Clunn & Kolia 1977), Sinaugoro avala-kavata and raba-avala, both labels for December and presumably meaning ‘start of avara’.

Some languages refer directly to rain, e.g. Tongan (Pn) vai-mua ‘early water, February-March’ and vai-mui ‘late rain, March-April’. Others refer to the rainy weather a little more obliquely, for example, Arosi (SES) hura-doa ‘blind month (because windy and rainy), April’ (hura ‘month’, doa ‘blind’); waru-ahe ‘all streams flooded, May’ (waru ‘stream’; ahe ‘flooded’); waro ŋaŋara ‘violent winds, August’ (waro ‘month’, ŋaŋara ‘rough (of weather)’.

The Sa’a (SES) term for a month around August is ro hutuhuto ‘the two foam’, obviously a reference to rough seas, but the allusion of ro ‘two’ in this context is not known.

One of two languages of communities that undertook long-distance voyaging refer to the odd month with a reference to this. Thus Motu veadi hiri-hiri ‘July’ refers to the hiri, the annual trading voyage westward to the Elema communities of the Papuan Gulf, powered by the southeast tradewind. Arosi ʔariha ‘January’ is a nominalisation of ʔari ‘come, go’ and denotes the month ‘when voyages are made’ at the beginning of the northwesterlies season.

Months in Yabem (Table 11.9) show poetic allusions to phenomena associated with the seasons. The numbers in the leftmost column are approximations to the lunar months denoted. The use of metonymic allusion is not restricted to Yabem. It is found in Mangap (NNG) (Table 12) and the Torres and Banks Islands languages (NCV) (Table 11.20). It was probably widespread, and perhaps many of the month names we cannot gloss are due to the fact speakers have forgotten their meanings or that researchers have not collected them.

4.8. Land crab migration⇫

Apart from the palolo, the only animal mentioned in month names across a range of languages is the land crab. Pawley (vol.4:173) writes:

The large land crabs, Cardisoma spp., are an important food source in many Oceanic communities and there are often several terms relating to their growth stage and spawning behaviour. For example, Foale (1998) reports that in west Gela (central Solomons) females of the abundant land crab Cardisoma hirtipes [Discoplax hirtipes] migrate to the beach at certain phases in the lunar cycle in the wet season, from October to December, releasing zoea larvae from their egg mass (lami). … Three weeks before doing this they migrate to the sea to immerse themselves, an event known as the sapa toga (thousands go seawards). This is the preferred time to take them because they are fatter. In Fiji, Wayans refer to the mass migration of Cardisoma hirtipes as vui.

Pawley notes elsewhere (2022) that spawning time of land crabs (tubā) in Waya ‘happens at high tide on two or three evenings in December, January and February’. Hence Wayan Fijian labels the season from December to February vula ni vūsē (vūsē ‘have or carry eggs or spawn’).

Panoff (1969:157) notes that in Maenge (NNG), the first month of the wet season (end of April to end of May) is called in some villages goga, the name given to ‘land crabs [which] leave the bush at night and gather on the beach where they are extensively caught by torch-light’.

Roviana (MM) has four month names, roughly October to February, that allude to the ɣarumu ‘land crab’: ɣarumu kara ‘month of the sea crabs’; ɣarumu leana (lea-na ‘good’) ‘[month] when crabs move to beach [to spawn – MR]’; porana hite (porana ‘poor condition’, hite ‘small’) ‘[month] when crabs are going off in condition, but some are still good; porana lavata (lavata ‘big’) ‘[month when] crabs are no good for food’ (Waterhouse 1949)

Fox (1955) defines Gela (SES) kakau (k.o. land crab) as ‘the name of the month of December when crabs come down to the sea to spawn’ while the following month, tivu popolo, lit. ‘look for [crabs] in covering’ refers to the time ‘when crabs return from the sea and hide in bushes’.

We have not been able to ascertain why the Maenge month name is attributed to May whilst the remainder centre on December.

5. Lunar ‘months’⇫

Lunar months were a useful way of planning for forthcoming festivals or intended trading voyages, or counting off the period of time during which a taboo applied to individuals following a birth or death. Ross quotes an elderly Takia (NNG) speaker describing the timing of a planned event (vol.2:288):

All right, and so they waited – in the old times they didn’t know about years. They always kept time by the moon. Thus, when they wanted to set a time – when they wanted to set a time, they mentioned the month. But they also didn’t know the names of the months. The moon waned and waxed, that’s all. They would say the months in this way: they would count the months with their hands, they would count them with their fingers. And then they would say, the month of the little finger will come and will die, the next finger will die, and the next and in the fourth months the man and woman will get married. They said this – well – with regard to their saying that they would marry in four months …

Ross saw in this a striking similarity with Whorf’s description of the Hopi conception of time. While westerners see time as if it is a physical entity, made up of measurable, countable units (minutes, hours, days of the week, months of the year), the Hopi do not make this extension. Ten days cannot be imaginatively experienced by the Hopi as if they were ten men. “[They] experience only one day, today; the other nine (or even all ten) are something conjured up from memory or imagination” (1956:139).

As Whorf describes it (1956:148):

The count [of days] is by ordinals. This is not the pattern of counting a number of different men or things, even though they appear successively, for, even then, they could gather into an assemblage. It is the pattern of counting successive reappearances of the same man or thing, incapable of forming an assemblage.

Like the Hopi, the elderly Takia speaker was counting successive visits of the same moon. Successive passings of cycles could be counted but they could not be aggregated into a countable unit forming a solid block of time. So although it is perfectly acceptable for Oceanic speakers to name different parts of a day (as when the sun is low in the sky) or a month (as when the moon appears as a slim sliver) or a year (as when the yams are sprouting), they did not add separate parts together to form a whole, or see a period of time as a whole that could be neatly subdivided. This conceptualisation applied to all astronomical cycles, whether daily, lunar or annual.

In spite of the elderly Takia speaker’s claim that “they didn’t know the names of the months. The moon waned and waxed, that’s all”, we have gathered lists of apparent lunar month names from more than thirty languages from Western Oceanic, the SE Solomons, Vanuatu, Micronesia, Fiji and Polynesia. These lists consist of, usually, 12 or 13 names for or references to natural features that mark roughly sequential points or periods of time in an annual cycle and reflect the complex conceptual system of astronomical, meteorological, ecological and horticultural checkpoints that made up the annual cycle in traditional Oceanic communities.

The lists are varied. While some listed names are direct references to or descriptions of events (see e.g. Sa’a, Table 11.19), others are largely untranslatable (e.g. Kilivila, Table 11.13; Nakanai, Table 11.17) . Some are strictly local references (e.g. Motu, Table 11.15). Still others refer to events through metaphors or allusions to shared narratives (e.g. Mangap-Mbula, Table 11.12; Yabem, Table 11.9). Some languages, like Sinaugoro (Table 11.16 in §11.5.2.2) and Mota (Table 11.4 in §11.4.4, Table 11.20 in §11.5.5) have many more month names than can fit in a year, and it seems that some or all months have alternative names depending on the checkpoints the speaker is focussed on. Indeed, the considerable variation in month names across Oceanic languages and even among closely related languages (e.g. Sinaugoro/Motu/Lala in §11.5.2.2; SE Solomonic in §11.5.4; northern Vanuatu in §11.5.5) suggests that alternative names have long been the norm.

It is significant that named natural features have a central focus but no precise beginnings and endings, so it is impossible to say when one period ends and another begins. There may be periods of overlap or gaps. This means that even if individual names for times of year were fixed, they could not be aggregated into a solar year. No doubt this explains the response of local speakers that they could not provide memorised lists when requested (in the Trobriands, §11.5.2.1.1, in Fiji, §11.5.6, in Tahiti, §11.5.8.1). Where neat sets of names exist, these are assumed to result from attempts to fit them into a western time frame

It is easy to read these lists of names as loosely comparable to our calendar months and assume because of their number that the names total a year. But this represents the intrusion of a western concept of time. Oceanic ways of talking about time may be compromised both by speakers’ own efforts to interweave the old system with the western one and by descriptions from ethnographers who have interpreted what they saw in terms of western concepts. The number of named lunar months in a list evidently had little or no place in Oceanic speakers’ conceptions of time. The fact that they are often organised into lists of 12 or 13 is a reflection of a western need to associate activities with western calendar months while also offering local speakers a link to a full moon or a lunar cycle if required. Theoretically, the number of names local speakers could use to identify points of time in a solar year is limited only by local knowledge.

In this light it is unsurprising that a set of POc month names cannot be reconstructed, even though it is reasonably clear from §11.4 which checkpoints were of importance to early Oceanic speakers. There is evidence, however, that in some places, related communities share similar month names, permitting low-level reconstructions. Examples include Loh, Mota and Mwotlap in the Banks Islands of northern Vanuatu, the Chuukic languages of Micronesia, and Polynesia. It needs to be stressed that the results cannot be regarded as a fixed system. Names cannot be combined to form a solar year. Lists will always include a degree of flexibility to ensure that descriptive names match natural features, and allow alternative names to be substituted as desired.

The following collections of month names are included with some discussion in an attempt to identify any systematic approaches.

5.1. North New Guinea⇫

5.1.1. Wogeo⇫

The Wogeo language is spoken on two of the Schouten Islands, Wogeo and Koil, located off the north coast of PNG. Table 11.10 shows Wogeo month names, pieced together from Hogbin (1938b:138–140). However, it is clear from Hogbin’s comments that the periods thus named may overlap.

Table 11.10 Wogeo month names

The ‘washing the Pleiades’ ceremony, mentioned in §11.4.2.1, was performed separately by each of four groups of villages. It is performed by the most northerly group in Bariat village in February “during the month when the Pleiades hang over this part of the island at sunset”, by a second group a month later “when the stars have moved round and the Pleiades are at sunset over Dap [village]”. A third group follows in April when the Pleiades are at Bagiau village, and a fourth group with their own month, at a time not noted by Hogbin. The months of, roughly, February, March and April are simply named the moons of Bariat, Dap and Bagiau. The moon of Bariat is also called kasawara (no gloss given). Where he does give the local names, Hogbin glosses some and comments on others. We have added one gloss, as rakum evidently reflects POc *rakum(u) ‘land crab’ (vol.4:173–174).

It is noteworthy that this is not primarily a horticultural calendar, as the staple crops, taro and bananas, are grown all year round, and so do not provide checkpoints. Instead, the calendar is based on various events in the natural environment. The months that refer to the Pleiades (§11.4.2.1), to the harvesting of canarium almonds (§11.4.5) and wasek nuts and to the southeast trade winds (§11.4.7) are anchored in the solar, not the lunar, year. Lunar-related events are the spawnings of (probably) the land crab (§11.4.8) and the palolo worm (§11.4.3).

5.1.2. Maenge⇫

Maenge (a dialect of Mengen, NNG) is located on the southeast coast of New Britain. Panoff’s (1969) listing of the months is shown in Table 11.11.¹⁸ The first gloss numbers the western months, but each Maenge month begins a few days earlier. This is the only calendar in our data in which almost every month name reflects the flowering of a wild plant (cf §11.4.6).

Panoff writes:

despite the continuous character of the process and the obvious impossibility of recognising clear demarkations between the stages, flowering always remains the final criterion and accounts for some calendar variations from village to village. Moreover, there are unavoidable discrepancies between time reckoned in lunar and in botanical terms whenever flowering happens to be late. (Panoff 1969:156)

Table 11.11 Maenge month names

This unit of time follows the phases of the moon very erratically. A vega matana ‘tree category’ ~ ‘month’ may last for as few as 15 days or for as long as five weeks. The Maenge turn their attention to the phases of the moon only to secure a correlation with the tree behaviour. Nevertheless the system is said to work satisfactorily with a margin of uncertainty of less than one month. Perhaps it is precisely the flexibility of the system which has allowed the Maenge not to add a thirteenth month to their calendar, as the observation of the shifting sunrise on the horizon through the year should have urged them to do if their monthly unit had been exclusively lunar (Panoff 1969:156).

The two seasons, vinte ‘the wet’ and kaepa ‘the dry’ (§11.3), each have six lunar months, and are separated by ulasi, the one month that has a fish name.

The Maenge are able to foretell palolo risings (§11.4.3) and also are aware of the appearance and disappearance of the Pleiades (§11.4.2.1), but Panoff does not recognise either of these events as a calendrical marker (1969:156, 158).

5.1.3. Mangap-Mbula⇫

The Bugenhagens’ listing of lunar month names in Mangap-Mbula (Vitiaz Straits:NNG) (2007b:420; Table 11.12) focusses almost solely on the annual cycle of the canarium almond crop (§11.4.5), although their accompanying notes are necessary to make that interpretation. For example, manᵐbule mamāza ‘the bird’s anus is dried up’ makes little sense without the accompanying note. ‘Galip’ is the Tok Pisin term for the canarium tree and almond. The ‘Word glosses’ represent our best attempt to gloss the component parts of the month names.

The one interpretable month name that refers less directly to canarium almonds is ⁿdām-bula ‘December’, where the first element reflects ⁿdāma ‘Pleiades’ (in the Marile dialect of Mangap this month name is simply ⁿdāma), but the Bugenhagens note that this marks budding of the canarium (and thus the beginning of the cycle).

Table 11.12 Mangap (Yangla dialect) month names

5.2. Papuan Tip⇫

5.2.1. Kilivila and Muyuw⇫

The calendars of closely related Kilivila and Muyuw provide a rare example of an apparently fixed list of names, to some extent divorced from the seasonal events reflected. Kilivila is the language of the Trobriand Islands. The Trobriand calendar has been the subject of considerable ethnographic debate (Malinowski 1927, 1935; Austen 1939; Leach 1950). There are two main areas of confusion. One centres on the number of months. The other is the difficulty of correlating these months with specific times of year in a place where four different calendars operate, and identifying the recognised standard by which the lunar months are kept in step with the solar year. The various descriptions of Trobriand lunar months by Malinowski and Austen throw up several unresolved issues, not only between the two authors, but also between Malinowski’s earlier and later papers. Leach’s paper is an attempt to resolve some of the differences.

5.2.1.1. The Trobriands

The Trobriands have a unique system, whereby four regions, Kiriwina (the major dialect of Kilivila) (in the north), Kuboma (centre), Kitava (east) and Vakuta (south), each run a distinct calendar with similar names and similar sequencing, but each starting their year with the milamala moon at a different time: Kitava roughly in June, Kuboma in July, Kiriwina in August and Vakuta in September. The moon called milamala is always the first (Malinowski 1935:463). The name milamala is also the name of the palolo worm (§11.4.3), which appears on the fringing reef of the Vakuta district during the moon of milamala ‘in September or even October’ (1935:54), presumably giving the moon its name. But the time of year to which it refers varies from one district to another. In other words, its sequence in a recognised list has taken precedence over its original association with a natural event.

Malinowski argues (1927:210) that the fact that each district starts its year at a different time is a reflection of different gardening systems. In places where taro is the staple food, gardens start early and are harvested early. In the main agricultural districts where small yams form the main crop, harvest occurs at least two months later than in the earliest yam districts.

The thirteen names, as given in Malinowski 1935 are: milamala, yakosi, yavatakulu, toliyavata, yavatam, gelivilavi, bulumaduku, kuluwotu, utokakana, (ilaybisila), (yakoki), (kaluwalasi), kuluwasasa. However, Malinowski (1927) and Austen (1939) are agreed that only ten names are significant. The three names whose existence is questioned (bracketed above) are those where informants are least certain, or are names recognised in one or two districts but not all. Two–yakoki and kaluwalasi—are listed by Malinowski in 1935 but not 1927. In 1927 he wrote (1927:215) that although

several moons seem to have different names in different districts, … it is significant that ten moons are identical everywhere and easily obtained, while the remaining odd moons, for which sometimes one to three names are obtained, always fall outside the scheme division of two groups of five.

Austen suggests that kuluwasasa is a corruption of kʷeluvāsasa, translated as ‘passageway between the [two] garden periods’ (1939:240). He writes that “the garden periods following ilaybisila are vague, and the moons really have no definite names and it is often a time of confused ideas” (1939:244).

Malinowski (1927:211) writes that milamala is the moon of festivities after the harvest, and that the sequence kuluwasasa, when harvesting is done, milamala and yakosi “are universally known to the natives and they are used by everybody”. Other month names figure less in people’s minds.

Leach (1950) offers an ingenious solution to the question of how this accords with a solar year. He suggests that the milamala period is the key, being a flexible period when taken as a whole across all four districts.

The milamala period of the whole group covers four months, this plus the other nine month names is sufficient to cover a full lunar year of 12-13 months. Thus in an ordinary 12-month year Kitava celebrates milamala nine months later than Vakuta, which is the ‘standard’, Kuboma one month after Kitava, Kiriwina one month after Kuboma, Vakuta one month after Kiriwina, which completes the annual cycle. The whole territory can thus complete a 12-month cycle without any one area bothering to count more than 10 months. So long as each group knows the relative position of its own calendar to that of its neighbour, the system is complete. … Vakuta milamala, as checked by the appearance of the palolo worm, is the beginning of a 12 or 13 month cycle which ends with the three months Kutava milamala, Kuboma milamala, Kiriwina milalmala (1950:254–255).

Austen (1939), however, sidesteps completely the difficulties of reconciling the palolo rising with the timing of the milamala month. Although his fieldwork in the Trobriands followed closely on the heels of Malinowski, he takes a position against Malinowski in arguing that the Trobrianders used a systematic series of solar observations in keeping the lunar months in step with the solar year. Austen believed that relations between month names, kʷeluva, and stars were more important than those between the month names and the moon. He writes that

there are quite a number of star groups (native constellations) connected with gardening, the most important being … Uluwa/Uruwa ‘Pleiades’ and Kibi ‘three stars of Aquila, the central one being Altair’.

In addition Austen lists sinata ‘part of Scorpio’, lakum (‘land crab’) ‘part of Cancer and Hydra incl. Praesepe’, dubukavivila ‘native constellation incl. Hamel’, kauwoma ‘native constellation incl. Aldebaran’, kiyadiga ‘Orion’s Belt’ and munukaiwau ‘Sirius’ (1939:240). And he adds later, “For the correction of the annual lunar cycle, the star group kibi is used” (1939:243). He notes that the community depended on inhabitants of Wawela, a village in the central Trobriands who were the accepted experts in knowledge of the stars, and who were consulted each year as to the timing of forthcoming rituals (1939:239).

Malinowski had written in 1927

that at certain seasons certain configurations of stars appear in the sky in the evenings. They have names for a number of constellations, for the Pleiades, for a part of Orion, the Southern Cross, and many others; and they know in which season these stars are visible, but they do not use them as a means of measuring time’ (Malinowski 1927:203).¹⁹ (our italics)

In 1935 Malinowski (1935:50–51)correlated moons, winds, and gardening and other activities with western months, but his chart contains no mention of stars. Austen’s evidence is discussed further together with Damon’s evidence for Muyuw.

5.2.1.2. Muyuw and the Trobriands

Damon (1982, 1990) describes the lunar calendar in Muyuw, spoken on Woodlark Island east of the Trobriands, and contrasts its system with that of the Trobriands, with which Muyuw is linked through the Kula Ring trading network. The calendars share similar names and month sequences, but differ in their starting positions and thus their relationship to the western calendar.

Table 11.13 contains 13 names of Kiriwina months (Malinowski 1935:51) and 13 names of Muyuw months (Damon 1990:290) aligned to highlight similarity of names (they are not aligned with reference to western month names). Nine of the thirteen show cognacy, evidence of a recent shared past. The names listed third, fourth and fifth contain a reflex of POc *apaRat ‘north-west wind’, the wind that holds sway from December to April. Apart from these three and milamala the names are, as far as we can tell, largely untranslatable and therefore without meaning-based association with particular seasonal events. Those pairs that are not cognates are marked with #, and occur close together in the sequence.

Although the two lists imply a shared system, Muyuw lacks a milamala moon. Oddly enough, the palolo worm appears in some Muyuw lagoons at the full moon of, usually, October, and it is called milamala. But it is neither eaten nor used for any calendrical purpose (Damon 1982:229).

However, there is, or was, another system of regulating time in Muyuw, one shared with the Trobriands, namely that provided by the stars. Damon (1990:29) provides a list of thirteen star groups and roughly aligns them with lunar months and other seasonal events or activities, although he warns: “Although stars are conceived to rise in a sequential order just as kwel [lunar months] follow one another, Muyuw do not associate particular stars with specific kwel.” (1990:38).²⁰

Several star names (Table 11.14) are cognate with Austen’s list for the Trobriands, although there are minor variations in their identity. Austen’s kibi refers to Aquila while Damon’s kib is Delphinus, an adjoining constellation. Strangely, although both list the Pleiades, their names for the star cluster are not cognate. In the Trobriands it is ulawa and in Muyuw gumeaw.

Damon (1990:38–40) implies that Muyuw rely on the stars rather than lunar months to arrange their yearly activities. He describes Muyuw knowledge of celestial bodies as impressive. “It is yam harvest time, people note, when the long axis of the Southern Cross stands vertical in the dark of a young night, mid-July or so”. And although other crops – taro, bananas, sweet potatoes – may be and are planted throughout the year: yams should be planted when gumeaw ’is thirty or so degrees above the western horizon at dusk, in February.”

5.2.2. Motu, Sinaugoro and Lala⇫

Motu, Sinaugoro and Lala are considered together as they are clustered in the middle of the segment of New Guinea’s south coast occupied by the Central Papuan group of Oceanic languages.

Lister-Turner & Clark’s (1954) Motu dictionary includes twelve month names, given in Table 11.15 with glosses drawn partly from Andrew Taylor’s (pers. comm.) notes and partly from Lister-Turner & Clark’s glosses.

Table 11.15 Motu month names

The Motu sailed their laɣatoi (double-hulled canoes) annually westward to the Gulf of Papua to trade with the Elema. They would wait for the mid-year wind change, then sail back to their villages (around present-day Port Moresby). These trading voyages were known as hiri, also used as the month name for a period roughly corrresponding to July. The names for January and October also seem to be associated with these voyages. The term maura (part of the month name for October) is glossed by Lister-Turner & Clark as

s.t. small given as a pledge, to remind recipient of his promise to return; a token that the messenger who brings it has been sent by the owner, an Elema man who sails on a laɣatoi and stays till next year.

The ‘promise to return’ probably alludes to a man’s pledge to his Elema trading partner to return the following year. ‘An Elema man who sails on a laɣatoi and stays till next year’ is almost self-explanatory: he is a man from the Gulf who is transported on the laɣatoi to its owner’s Motu village, and remains there as a guest until he can return to the Gulf on next year’s hiri.

Sinaugoro, to the east of Motu, shares some month names with Motu, but none that are associated with the hiri, in which Sinaugoro speakers did not traditionally participate. Instead, names peculiar to Sinaugoro often reflect the fact that its speakers live in the savanna. Whether the names shared with Motu are loans or shared inheritances is difficult to know, as the languages are phonologically similar. Table 11.16 is drawn from two sources, Kolia (1975) and Tauberschmidt (2007). Kolia’s data are from the Balawaia dialect. He does not assign his month names to western calendar months, but does gloss them. Tauberschmidt on the other hand, whose month names are from the Saroa dialect, assigns them to western months, but does not gloss them further. Some glosses, marked (R) (= Ross), are inferred from Kolia’s and Tauberschmidt’s vocabularies.

Table 11.16 Sinaugoro month names (Balawaia dialect after Kolia 1975; Saroa dialect after Tauberschmidt 2007)

The range of names in Table 11.16 demonstrates that month names even in dialects of the same language can diverge quite widely. It also reveals how culture-specific month names can be. Where some Motu names reference seafaring, names from the Balawaia dialect reference a grass species and burning in the surrounding savanna. Both dialects refer to the end of the harvest in April and the beginning of a new crop in May.

Both dialects also include manu-bada, perhaps the ‘big bird’ constellation of Sirius, Canopus, Procyon, Betelgeuse and Rigel described in vol.2(168–170), which Tauberschmidt assigns to January. Immediately after sunset in January this huge constellation dominates the eastern sky over the Motu and Sinaugoro villages. That ‘big’ has its Motu shape bada rather than Sinaugoro bara suggests that it may be a loan.

Clunn and Kolia (1977:143) list thirteen month names for closely related Lala, situated to the west of Motu. The name for the thirteenth Lala month, avala, is literally ‘north-west’, reflecting POc *apaRat ‘northwest wind; wet season when westerlies blow and sea is rough’ (vol.2:129). This is cognate with Sinaugoro avala, occurring in two month names. Otherwise no Lala month name is cognate with either a Motu or a Sinaugoro month name.

5.3. Meso-Melanesian⇫

5.3.1. Nakanai⇫

Chowning (2016) lists eight lunar month names from Nakanai, on the north coast of New Britain. The eight are considered a complete set, and their ordering is clear, running from e tolo bubu ‘1st moon of the calendar year, approx. Sep.–Oct. (starts about autumn equinox)’ to e sakalu kea (lit. ‘white reef’) ‘8th and last moon—actually several moons of the same name, approx. June–Sep.’ Apart from the 8th month, the names are, as far as we can tell, untranslatable, and there is no discussion apart from brief descriptions included in the dictionary entries.

Table 11.17 Nakanai lunar months (Chowning & Goodenough 2016)

5.3.2. Barok⇫

Wagner (1986) describes the timing of the gardening cycle in Barok, New Ireland. He writes that “a cycle of six named lunations seems to have been recognised, and used in conjunction with gardening, in pre-European days.” (1986:38). His own observations and his conversations with Barok speakers about the traditional calendar lead him to conclude that the awat ni nien ‘season of plenty’ and the awat nere loŋ ‘season of hunger’ each represent a six-month cycle, but because the large garden is planted only four months after the small (see Table 11.18), the gardening calendar cannot be described within a single six-month cycle. He points out that the Barok observe lunar, solar and sidereal events (1986:41), which do not correspond exactly. However, he concludes that each awat effectively begins at the new moon closest to when the Pleiades are crossing the meridian, i.e. in the evening on 21 February and predawn on 22 August. The position of the sun around these dates is also not far from its position at the solstice.

Table 11.18 is the table given by Wagner but augmented from the text (1986:36–44). Since the Barok reckoned time in terms of six-month seasons, the same set of six months occurs in each awat, i.e. twice in a year, a unit that the Barok were allegedly unaware of. The blanks in the “Gardening” column of the table are due partly to the fact that Wagner’s table includes them, but partly to the fact that Wagner does not mention the harvests. The numbers in the leftmost column represent the month whose second half falls into this lunation, e.g. 01 represents the latter part of January and the earlier part of February.

We note that this account contains one oddity: since the months in each awat have the same names, tege ni kuka ‘moon of crabs’ occurs twice a year. In reality, however, land crab spawning (§ 11.4.8) occurs only once a year, presumably in November. This leads to the speculations that the Barok were aware of the annual cycle (as the distribution of garden planting described by Wagner also implies) and that the six-month cycle represents a reduction of a set of terms that once covered the year.

Particular importance is attached by the Barok to the Pleiades. In one marana-kai they are crossing the meridian at dusk on 21 February, but have already been prominent in the evening sky for about a month. In the other marana-kai they are crossing the meridian at dawn on 22 August, and will remain prominent in the pre-dawn sky for about a month.²¹ This account is puzzling, as Western Oceanic speakers typically make no use of the meridian (it is not clear how they would identify it). A possibility is that Barok speakers observed the risings and settings of the Pleiades. Their apparent first pre-dawn rising around 7 June would very roughly fit the beginning of awat nere loŋ ‘season of hunger’, and their apparent pre-dawn setting around 5 November would be a herald of the monsoon and the awat ni nien ‘season of plenty’.

5.4. Southeast Solomons⇫

Sets of lunar months have been recorded for Gela (Fox 1955), To’aba’ita (Lichtenberk 2008), Sa’a (Ivens 1918, 1927), Kwaio (Keesing 1975) and Arosi (Fox 1978). Other dictionaries including ’Are’are (Geerts 1970) and Owa (Mellow 2014) contain partial lists. Except for Gela, these languages belong to the Malaita-Makira subgroup of SES.

Table 11.19 Sa’a month names (adapted from Ivens 1927)

The most comprehensive of these lists is Ivens’ (1927:396-397) for Sa’a, shown in Table 11.19, which varies in some respects from that in Ivens (1918). In 1927 he describes the year as “beginning with the flowers of the canarium almond in September (hule i lade ‘flowers arrive’) and ending with the period of ripe almonds (ŋali maelo) in August”.²² August is also called ro hutohuto ‘froth, foam’, referring to the month of wild weather and big seas. (In Arosi the same month is referred to as waro ŋavara ‘rough weather month’.) Four months – October to December – are prefaced by oku, the palolo worm, while three – April to June – are prefaced by loʔa (meaning unknown, but used also as month name prefix in Lau (loa) and possibly in To’aba’ita (loʔi). Ivens describes the loʔa months as relating to ‘the yam harvest which is dug about the beginning of May’.

A comparison of month names across languages shows broad agreement in Gela, Sa’a, Arosi and ’Are’are with respect to the palolo months, although both Sa’a and ’Are’are give the name for the palolo to four months, suggesting it now means something like ‘month, season’. SES palolo month names are listed in §11.4.3.

5.5. Vanuatu⇫

Unusually detailed lists of month names are available for the Torres and Banks islands of northern Vanuatu, thanks to various anthropological studies (Codrington 1891 for Mota, Durrad 1940 and Mondragón 2004 for the Loh dialect of Lo-Toga, François 2023 for Mwotlap, and François pers. comm. for Lo, Mota, Mwotlap and partial Mwesen). They are summarised in Table 11.20.²³

In the leftmost column of Table 11.20 numbers approximate western months. Terms for October, November and December in Loh, Mota and Mwotlap make reference to the small and big risings of the palolo (§11.4.3). Loh appears to be out of sync with Mota and Mwotlap by a month, but this probably reflects the vague relationship between western months and the lunar months labelled here.

According to Durrad, five Loh months, from April to August, are named for the seasonal Panax grass called moɣot, which springs up as winter ends (cf §11.4.6). In April the grass is fresh (meta ‘wet’) and in May it dries off (reŋ ‘dry’). The name continues to be associated with the next three months—becoming a seasonal name now associated with digging up first the yam-like təmeɣ in June and then the kʷɛtə in July. A further five months, from November to March, are named for the palolo worm, n’ut. As with moɣot the name continues beyond the true palolo months, becoming a seasonal name. Durrad also links each month with gardening activity, and at times, as in June and July, this is reflected in the month name. As planting is carried out in September and October, the yam vines are in full green in February and turning rusty red in March, giving rise to the terms for green and red in their respective month names. Yam harvesting occurs in May, the təmeɣ [wild yam] and kʷɛtə mentioned above reaching maturity later than cultivated yams.

In Mota, as there are many more month names than can be fitted into a lunar year, the list given is a selection among the possible names. Included are three maɣoto months, May to July, two rara months, August and September, and three palolo (un) months, October to December. The maɣoto month of May is translated as ‘fresh grass’, equivalent with Loh ‘wet grass’. Codrington (1891:350) writes that, for Mota, maɣoto and rara, the erythrina or coral tree (from POc *rarap, vol.3:158, 257) have become seasonal terms for summer and winter respectively, roughly equating with the annual division of the Loh calendar (cf §11.4.6).

Table 11.20 # FIXME: Table 11.20 not legible in PDF

The Mota, Mwotlap and partial Mwesen calendars display more similarity with each other than any of them does with Loh. The three make similar references, even when the terms are not cognate. The terms for April, however, are cognate, reflecting what François (pers. comm.) reconstructs as Proto Banks *lamʷas-aki dorodoro, apparently a serial verb construction consisting of *lamʷas- ‘to hit, strike’, the applicative suffix *-aki, hence ‘strike against’, and dorodoro ‘to rattle’. The meaning of the construction is evidently ‘to strike against so that it rattles’, the agent being the wind, the patient the reed-like grass Miscanthus floridulus which makes a rattling sound when the dry reeds, blown by the wind, strike each other (François, pers. comm.).

The terms for January, February and March in Mota, Mwotlap and Mwesen all make oblique reference to the (monsoon) winds, in Mwotlap and Mwesen striking the reeds, and in Mota, vusi-aru [beats casuarina], where the wind instead strikes the casuarina (aru). The terms for January refer to the reeds thickening (Mwesen revsos), budding (Mwotlap ni-witɣoy) and flowering (Mota wotɣoro). In February the reeds are preparing to expel their seeds. In Mwesen they ristek taqan ‘grab hold’ (of their bellies, like pregnant women), and in Mwotlap they towowoh ‘burst open’. In March the wind blows so hard that in Mota fragments of the reeds break off and fly away (tete), while in Mwesen it ‘blows off white hair’ (whether the hair belongs to the reeds or the old men is uncertain). François has no gloss for the Mwotlap terms, but it is probable that tit is cognate with Mota tete ‘fly away’.

Lest the explanation for the abbreviated sentences used as month names here appear too imaginative, it is worth noting that Codrington & Palmer’s dictionary entries tell the same story as François recorded from Mwotlap and Mwesen speakers.

Codrington (1891:349) warns that in Mota

it is impossible to fit the native succession of moons into a solar year; months have their names from what is done and what happens when the moon appears and while it lasts; the same moon has different names. If all the names of moons in use in one language were set in order the periods of time would overlap, and the native year would be artificially made up of twenty or thirty months.

The reason for Codrington’s warning is that Mota months were named not only for the palolo and the effect of the seasons on wild plants as in Table 11.8 but also for periods in the yam cycle. It would be possible to list month names in order from April to December that referred to events from preparing the yam garden to planting, maturing, harvesting and clearing.

Because there are some cognate forms among these languages, it is possible to reconstruct three lunar month names for Proto Torres-Banks, using Clark’s (2009) orthography, and partially reconstruct two others.²⁴

Lists of names are available to us from three Malakula languages: Atchin, Naman and Avava. However, these display almost no cognate forms either with each other or with the Torres-Banks languages. The exceptions are Atchin ul wele ‘little palolo’ and ul lɛp ‘great palolo, October–November’. Moreover, Capell & Layard (1980) contains nineteen Atchin ‘month’ names, whilst in Avava the year is divided into just eight periods. Clearly, these are not strictly lunar month terms.

Also available are lists of month names from southern Vanuatu: Sye, Kwamera, Lenakel and Anejom. However, apart from a brief discussion of Sye names by Crowley (1998), the names are compiled from dictionaries without ethnographic description or interpretation of their meanings. There are almost no cognates, even between Kwamera and Lenakel, the two languages from Tanna island.

5.6. Fiji⇫

Hale (1846:68) offers a list of twelve names for Fijian lunar months obtained from missionary sources, with two (February and March) referring to flowering of the ŋasau reed (Miscanthus floridulus), three (April-June) referring to garden cultivation, two (October, November) to the palolo and two (December, January) to the nuga, a rabbitfish (Siganus vermiculatus). However, he comments

Besides the appearance of the mbalolo, the natives have few means of determining with exactness the progress of time. Indeed, they pay little attention to this, and we were unable to obtain from several to whom we applied, the names of the months in their regular series. … The Feejeeans know nothing of astronomy, and have not even names for the most important constellations.

Seemann (1862:296-299) refers more confidently to “the eleven months into which the calendar is divided”, quoting “an intelligent Bauan chief and the consular interpreter.” He comments that “the names given by me, as well as their succession, do not quite agree with those given by Wilkes (i.e. Hale from the Wilkes expedition). The names of the months may also be different in different parts of the group”. However, the list in Table 11.21 from Seemann, including his comments, substantially mirrors the horticultural/faunal/floral annual cycle, and accords with later broad descriptions such as those of Hocart (1929) for the Lau Islands of eastern Fiji and Pawley and Sayaba (2022) for Wayan, spoken in western Fiji.²⁵

Table 11.21 Bauan Fijian month names according to Seeman (1862)

He notes that Hazlewood, who published a Fijian–English dictionary in 1850, allowed four months, from May to August, to be effectively the clearing month. Presumably this provided the degree of flexibility necessary so that the more precise seasonal events such as the rising of the palolo occur at the expected time.

5.7. Micronesia⇫

In passing from Fiji to Micronesia we encounter a change in the way the year is envisioned. In Micronesia, unlike the other areas examined, there is internal consistency in that stars, and stars alone, are seen to mark the passing of time. All month names reconstructed are star or constellation names (vol.2:166–184), reflecting a calendar with a smooth transition from one star or star cluster to the next. Alkire (1970:37–38) writes that in Woleai “the seasons and their subdivision of months or moons (maram) are initiated by the rising and setting of designated stars.” The months that define the winter (approximately November to April) and summer (approximately May to October) seasons are all named after the star or star grouping which heralds the beginning of the period (1970:38). As names of months the Micronesian terms are strictly speaking only reconstructable for PChk, based on month names for Truk (= Chuuk) (Goodenough & Sugita 1990), Mortlock and Lamotrek (Christian 1899), Ponape (Kubary 1895; Christian 1899), Woleai (Alkire 1970), Carolinian (Jackson & Marck 1991), Puluwat (Elbert 1972) and Sonsorol (Capell 1969). However, some of the terms have non-Chuukic cognates within Micronesia and are reconstructable at least as star or constellation names as far back as PMic. PChk reconstructions are given here in the same orthography as PMic, that of Jackson (1986). Terms from Mortlock and Lamotrek are given in the orthographies of their sources, as we have insufficient information to convert them to the orthography otherwise used in this volume.

Speakers from the Polynesian outlier island of Kapingamarangi, which lies roughly midway between Mussau and the Caroline Islands of Micronesia, have largely adopted the Chuukic calendar, ten of their twelve months according in order with the Micronesian listing (Elbert 1948). These are treated as borrowings. A single Kapingamarangi month, matariki, accords with a Polynesian term.

Sources for Puluwat and Carolinian each list twelve names and link them to western months; those for Mortlock, Lamotrek and Sonsorol list months in regular sequence from one to twelve. The Trukese Dictionary lists fourteen names as months in the traditional sidereal calendar, only four of which are numerically ordered. An additional one is listed as ‘named for a month in some calendars’. Some of the cognates in the sets listed below include star names that are not month names. Sixteen PChk terms have been reconstructed that, taken in sequence, albeit with some overlap, represent a solar year. Of these, four (Leo, Corona Borealis, Vega and Andromeda) are represented as month names in only two or three languages, and are regarded as probably substitutions for other nearby stars or star clusters.

Because each year the new moon rises eleven days earlier than in the previous year, the time of year thus delineated moves between two adjacent western months.²⁶

We have no clear evidence that any star or star cluster is regarded in the Chuukic group as heralding a new year. Christian begins his list with Leo for Mortlock and Corvus for Lamotrek, while Capell also starts with Corvus for Sonsorol, i.e. roughly September or October, but neither author discusses his choice. In Woleai the star Arcturus, identified with November, is identified as marking the change of seasons from summer to winter while Pegasus, identified with May, gives its name to the summer season. Because of its dominant role elsewhere in the Oceanic world we begin our list with the Pleiades.

It is tempting to try to associate this with PPn *mataliki ‘Pleiades’ (lit. ‘small eyes’), but this is apparently a chance resemblance.

There is some overlap in associating the next two reconstructions with roughly the same month. Although Lamotrek and Sonsorol attribute terms to two successive months, Mortlock combines both for the same month. Christian (1899:393) does, however, offer alternative Mortlock terms. Aldebaran and Orion’s Belt are close together in the night sky, and both could be identified with the same time of year.

Although the next set takes its name from the giant Bird constellation, which covers a considerable area of the sky, it is the appearance of its brightest star, Sirius, that is most consistently identified with the moon rising in September.

According to Christian (1899:389), Lamotrek sorabol is derived from sor ‘look’ and bol ‘taro patch’ and means ‘viewer of the taro patches’ because it shines during the taro season.

Lamotrek aramoi is allegedly derived from ara ‘conclude’ and moi ‘come’, as the ascendancy of Arcturus marks the end of the NE winds that bring visiting parties to the island (Christian 1899:389).

It is tempting to associate the next reconstruction with PPn *mataliki ‘Pleiades’ (lit. ‘small eyes’), but the similarity is evidently accidental.

Christian (1899) attributes this constellation name in Mortlock to stars in Hercules, but the star chart suggests that Abo et al (1976) are correct in assigning it to stars in Sagittarius, as there is nowhere near a month’s difference between Antares (the previous month) and Hercules.

The next reconstruction, literally ‘big mati’ contrasts with PMic *mati-ciki above, ‘small mati’ (meaning of mati unknown). Elbert (1972) attributes the meaning ‘big/old breadfruit’ to the Puluwat term, since the Puluwat breadfruit season lasts from May until December (mæy > PROc *maRi ‘breadfruit’). However, there is clear support for PMic *mati- as the first element.

Christian (1899:394) records Yapese orthographic lagu (probably laxu) for a month around June. If this is an early borrowing from a Micronesian language, the PMic form was probably *laku.

PChk *kua below evidently reflects PMic *kua ‘Dolphin constellation, a constellation including Cassiopeia and approximately equivalent to Aries’ (vol.2:170).

The PChk reconstructions in Table 11.22 show a smooth progression of rising stars over a year. Carolinian, Chuukese, Mortlock, Puluwat, Woleaian, Lamotrek and Sonsorol follow this sequence closely, each containing reflexes of 12 of the 16 reconstructions, although varying slightly as to which are not represented. The four star clusters marked with an asterisk are those listed as calendar stars in only two or three languages and are probably alternatives for other close stars representing the same period.

The only Micronesian language outside western Micronesia (Marshallese + Chuukic) in which terms for divisions of a year have been located is Kiribati, which has a rather different sidereal calendar. Grimble (1931) provides a detailed account of Gilbertese [Kiribati] astronomy. The year is considered to begin with the appearance of the Pleiades about 15 degrees above the eastern horizon just after sunset, in about the first week of December. The Kiribati uniquely use the rafters of the meeting house as a grid reference by which they locate their stars. Each six-monthly season is subdivided into eight periods called boŋ, measured by the successive altitudes of the seasonal star at the hour after sunset as observed through the grid of the rafters of the meeting house (1931:200). These sixteen named divisions, each a little over three weeks long, subvert any need for a calendar based on lunar months.

5.8. Polynesia⇫

5.8.1. Organisation of the lunar calendar⇫

Lunar calendars are ubiquitous in Polynesia, or at least were, prior to the introduction of the western calendar. Names commonly recur, theoretically permitting reconstruction to PPn, but comparison of lists shows immediate problems. Makemson (1941:97–98) has noted that different islands of the Hawaiian group use more or less the same month names, but in different orders. Not only do starting months differ, but the order of months is often scrambled. The same is true of lists from Tahiti, Samoa, Tokelau and the Marquesas. Some calendars consist of twelve months, others thirteen, while some, including Tonga, Manihiki, Tahiti and Hawaii, have a thirteenth month interspersed occasionally. East Futuna has fourteen months recorded but this apparently occurs because two months, December and January, each have two names, presumed alternatives, listed. Kirch and Green (2001:310) comment that

since … the original calendar was strongly correlated with local ecological conditions and with yam horticulture, it is not surprising that the calendar was significantly reorganised once people had left the homeland region, and once their ecological settings and horticultural practices had changed.

A further explanation for these discrepancies comes from Gill (1876:317) in his description of lunar months on Mangaia in the southern Cook Islands:

The knowledge of the calendar belonged to the kings, as they alone fixed the feasts in honour of the gods, and all public spectacles. For others to dare to keep the calendar was a sin against the gods, to be punished by hydrocele [drowning?].

This perhaps explains why many lists have been recorded with a degree of doubt by the informant as to their veracity. Even the earliest records show a degree of uncertainty, as in the following 1789 account from Bligh who sought lunar month lists from Tahiti (quoted by Oliver 1974:268):

To get a certain Knowledge of their division of time has given me much trouble, for altho many people pretend to know it, Yet I have found them so contradictory in their Accounts as convinced me they were not acquainted with the particulars of it. Tynah assured me only a few Old People could give me any information and that he knew it but very imperfectly himself.

For practical purposes the account here relies on Williamson (1933) and Collacott (1922) for Tongan, Rensch (1986) (based on Grézel 1878) for East Futunan, Williamson (1933) (based on Turner 1884) for Samoan, Besnier (1981) (based on Kennedy 1931) for Tuvalu, Makemson (1941) for Manihiki, Williamson (1933) for Tokelauan, Gill (1876) for Mangaia, Makemson (1941) for Penrhyn, Oliver (1974) for Tahiti, and Makemson (1941) for Hawaii.

Comparison of these lists shows frequent recurrence of lunar month names, but some diversity in the month attributed to each name. While reconstruction of the form has been possible, its position in a sequence has proved arguable. Twenty one reconstructions of the name (but not the associated month) have been made, with 8 from Proto Polynesian, 11 from Proto Nuclear Polynesian and 2 from Proto Ellicean. There are a number of pairs, identified by muqa ‘in front’ vs muri ‘behind, last’ (or toe ‘again’) and others by contrasting terms kelekele ‘dirty’ vs maqa ‘clean’ and maquri ‘alive’ vs mate ‘dead’. Their cognate sets appear below in §11.5.8.3.

A major clue in the ordering of putative PPn lunar months lies in the fact that a number of languages follow similar sequencing for up to four or five month names, as shown in Table 11.23.

The table shows the assignment to months (January = 1 etc.) of reflexes of reconstructed lunar month names in Tongan, East Futunan, Samoan, Tuvalu, Manihiki, Tokelau, Penrhyn, Tahiti, and Hawaii. It can be shown that East Futunan, Samoan, Tuvaluan, Manihiki, Tokelau and Tahiti share some sequences. These are genealogically and geographically quite widely scattered languages, and their sequencing can be taken as a good indication that the system in PPn was similarly ordered.

The difficulty then becomes one of arranging the preferred sequencing into the appropriate times of the calendar year. Names could be expected to relate to horticultural processes, particularly in parts of western Polynesia to the yam crop, or in parts of eastern Polynesia to the breadfruit season, and to the wet-dry seasons, while links to the palolo worm are prominent. There are also links to significant stars or star groups. In Futuna, for example, successive months were marked by the Pleiades, Orion’s Belt, Sirius, Regulus and a group of stars recorded as possibly the Southern Cross. In Hawaii successive months were marked by the Pleiades, Betelgeuse, Sirius and other stars whose names were not recorded. In particular, however, it is the appearance and disappearance of the Pleiades that plays a significant role in the determination of the annual calendar.

5.8.2. Reconstructions⇫

PPn *mataliki is reconstructable for ‘Pleiades’ (vol.2:165), but as a month name only in PNPn (not in Tongic), where it is marked by the new moon after the first pre-dawn rising of the Pleiades which occurred in 500 BC in mid-May, so here accorded to June. As Kirch and Green (2001:262, Table 9.4) have noted, the risings and settings of the Pleiades were widely observed in many Polynesian societies, “where they were used to mark the change in seasons and/or to mark the commencement of the year.”

Since the internal subgrouping of Polynesian languages is relevant here, the entries below are marked either ‘Tongic’ or ‘Nuclear Polynesian’, the two first-order subgroups within Polynesian.

PNPn *[kau]unu-unu is reconstructed for the next term, rather than *kaununu. It immediately suggests a link with PChk *ūnu ‘late July, early August; the star Aldebaran’, and an ordering following the Pleiades and preceding Orion’s Belt as in Micronesia. The PPn prefix *kau- had two functions: to derive a collective noun meaning a ‘group, company, bunch of s.t.’, and to derive an instrumental noun, generally a long thin object (here PPn *kau < POc *kayu ‘tree’). The Samoan and Tahitian reflexes lack an expected initial ʔ-.

Strictly speaking, the next item should not be reconstructed, as it is reflected in only one language. However, it is noted here because it is a self-evident case of a (part-)constellation marking a month. Orion’s Belt is a salient three-star feature in the sky. The middle star, Alnilam, has its pre-dawn rising in late May, i.e. about two weeks after the Pleiades and a month before Sirius, which occurs as a month name for July. Hence the attribution of *tolu to June. Note also PChk *tolu-tolu ‘about August; three stars of Orion’s Belt’.

Kirch and Green assign the two *siliŋa months to January and February, whereas here they are assigned to June and July. Their reasons for the former assignment are that Outlier and East Polynesian languages place them around January and February and that the second word of *siliŋa kelekele seems to be cognate with Fijian kelikeli ‘March’. However, neither of these grounds holds strongly. Both first-order subgroups of Polynesian, To and NPn, have reflexes pointing towards June and July, whereas the January/February reflexes occur only in NPn, specifically in Outlier and East Polynesian languages which are unreliable witnesses because of their lengthy and sometimes tortuous migration histories. Tongan evidence is weighted more strongly because it lies in the Polynesian homeland. PPn *kelekele here means ‘dirty’, in contrast with *maqa ‘clean’, and is cognate with Fijian gelegele ‘dirty’, not with kelikeli ‘a hole dug, a ditch’ or kele ‘pile up (yams)’.

Unfortunately, whilst the meanings of *kelekele and *maqa are clear, the meaning of PPn *siliŋa is not: it is probably a nominalisation of either PPn *sili ‘exceed’ or PPn *sili ‘put on top of’ (Biggs & Clark 1993). It is possible that it is derived from the latter and denotes stages of garden preparation. The Tuvaluan reflexes have each lost one word of the two-word phrase, while Hawaiian has replaced hilina with hinaia (a word which crops up in several month and star names). Penrhyn has adopted the pattern of other month pairs and replaced *kelekele and *maqa with reflexes of *muqa and *muri. Each of these changes probably reflects the fact that speakers no longer knew the meanings of the terms.

PPn *takulua is readily reconstructed as the name of a bright star (vol.2:163), and scattered reflexes in Eastern Polynesian languages indicate that this was Sirius. The NPn languages listed below are all Eastern Polynesian. It is assigned here to July on the grounds that its pre-dawn rising at Niuatoputapu occurred in late June, but this may be to project too much back to the PPn stage. In Micronesia the star Sirius is linked with September.

We assign PNPn *oloamanu to August (as do Kirch and Green) on the basis of its ordering relative to other months, and especially of the fact that it precedes *palolo muqa, which cannot be assigned earlier than September. The East Polynesian data (Manihiki, Penrhyn, Marquesan and Tongarevan) are unreliable because of their lengthy and tortuous migration histories.

In parts of Polynesia, just as in parts of western Melanesia, the palolo worm (Eunice viridis) spawns in a spectacular and predictable manner during October/November and November/ December during the last quarter of the moon. It is a well known phenomenon in Tonga and Samoa, as it is in Fiji, but, as far as we can tell, either does not occur or is not recognised in other parts of Polynesia. However, the name for the worm is known in Tuvalu and East Uvean at least. In East Futuna palolo-muʔa and palolo-muli refer to star names and thus the lunar months identified with these stars. In Tokelau, Mangaia, Tuamotu, Tahiti and no doubt other parts, the terms exist apparently only as month or seasonal names.²⁸ Lack of Tongic reflexes means we cannot reconstruct these terms to PPn in spite of the fact that reflexes occur in compounds in well separated languages. PNPn *palolo muqa and *palolo muli referred to the minor and major spawnings of the palolo. More than any of the other month names, we can be reasonably sure of the chronological assignment of the palolo months because of the regular pattern of the palolo’s spawning. Their association with the wrong months, even in Samoa, indicates that they have become primarily names in a system divorced from their original meaning.

Kirch and Green (2001:268– 271) cite the Bauan Fijian terms balolo lailai ‘small balolo’ and balolo levu ‘big balolo’, respectively October and November, as external evidence for the PPn terms. Although the words for ‘small’ and ‘big’ in Fijian have been replaced in PPn by ‘be first’ and ‘be last’, this is a legitimate inference. The durability of these month names can perhaps be attributed to the co-occurrence of three events: the major palolo spawning, the last post-dusk setting of the Pleiades, and the beginning of the wet season and the season of abundance. Together, these events made *palolo muli a highly marked time in the Polynesian annual cycle.

Kirch and Green assign PNPn *munifa to December. Although the meanings of its reflexes do not give much support to this reconstructed meaning, they seem to be correct, as the month denoted by its reflexes immediately follows *palolo muri in E. Futunan, Samoa, Tokelau, Tuvalu, Manihiki and Tahiti. The term may be more correctly muri-afā ‘end of storms’ (< POc *muri + *apaRat), referring to the end of the trade-wind season. Tuvaluan murifa seems to represent a halfway stage en route from PNPn *munifa to PEPn *muriafa.

PNPn *takaoŋa is attributed to January because it follows the Tuvaluan and Penrhyn month whose name reflects *muri(a)fa. Makemson (1941:214) and Pukui and Elbert (1971) both write that it is the name of a star in Hawaii.

The next two month names contain the term *lisa ‘louse’s egg, nit’. Kirch and Green (2001:271) attribute them to July and August. They cite Collocott (1922:167), who explains that little protuberances (“nits”) appear on the seed yams during these months as they are forming roots. PPn *muqa and *muri respectively mean ‘be first’ and ‘be last’. Month names are instead attributed here to December and January on the basis of the glosses of their reflexes. They are at best weakly supported reconstructions.

The reconstruction of PNPn *utua muqa ‘January’ and PNPn *utua muli/*toe utua ‘February’ is fairly straightforward, as the meanings of their component words are clear and match the horticultural season in which their glosses place them. PPn *utu meant ‘harvest yams’, its nominal form utua referring to ‘yam harvest’.

PNPn *tākelo ‘Orion’s Belt or Betelgeuse; January’ is attributed to January on a combination of clues. It appears to have referred to stars in the constellation of Orion. Whereas *tolu above evidently marked their pre-dawn rising in June (as does the Marquesan reflex of *tākelo), the glosses of the Samoan (‘January’) and Nukuria (‘February’) reflexes together with a Samoan comment that this is ‘a month of wind and storms’ (Henry 1928:234), imply that *tākelo referred to the post-dusk rising of the stars in Orion’s Belt in December.

Kirch and Green interpret PPn *wai muqa and *wai muri, where *wai is ‘fresh water’ or ‘rain water’, as referring to the months at the end of the wet season. The three Sa’a (SES) months which make reference to wai ‘fresh water’: Feb. loʔa wai mʷai-mʷai ‘small water’, March loʔa wai paine ‘big water’ and April loʔa wai (meaning of loʔa unknown) also fall in the same period (Ivens 1927).

Kirch and Green’s (2001:272) reconstructions for the next two months contain the word *faka-qafu ‘cause to be heaped up’ (*faka- ‘causative’ + *qafu ‘[be a] heap’). They take this to denote the preparation of gardens for yam planting at the beginning of the dry season. However, the Tongan and East Futunan reflexes point to PPn *faka-afu (*afu ‘shoot or sucker’).

It seems that the Polynesians were more systematic than speakers in western Melanesia in tying names specifically to lunar months rather than just to times of year. Although we have established a more or less sequential order of month names, we need to be cautious. Perhaps we need to be reminded here that a) the idea of a year as a fixed period of time did not exist for early Oceanic speakers (note Williams’ Maori definition of tau as ‘season, year, the recurring cycle being the predominating idea rather than the definite time measurement’), and b) times identified within a year were identified not by their length but rather by their central focus. It is easy but mistaken to read the lists through western eyes, and interpret them as fully equivalent to western calendar months. The interpretation is reinforced because we have little option other than to translate local terms for time of year either with western month names or by numbering lunar months from 1 to 12, with an arbitrary starting point. But there was no attempt by local speakers to fit the lunar months into a fixed period of time. The lists were not fixed; rather they were ready lists of more or less sequential markers of time within an annual cycle that could be moved or added to if moons became out of step with natural features.

<