Ethnology of Manihiki and Rakahanga
Though a low-lying coral atoll does not supply suitable basaltic material, a good deal can be done with coral boulders and coral limestone slabs. Rakahanga and Manihiki, however, do not show the activities in available material that characterize Tongareva. The outstanding Tongarevan features, house sites and religious inclosures, are lacking, and stone piers for canoes were not observed.
House sites. The stone platform, according to my informants, was not used. The surviving native house on Rakahanga is characterized by the absence of any platform or stones marking the wall boundaries. This is in marked contrast to the platforms of coral boulders associated with the Rarotongan types of houses which have become popular. The platform was introduced with the type of house. Because the present modern villages occupy the sites of the old villages, any traces of the past that might have survived the changes of time have been obliterated. The custom of segregation within the present village sites from the earliest times prevented the establishment of houses on other islands that might have left traces of interest to the archaeologist. It seems true, however, that not only were no house platforms made formerly, but that even the marking of the wall boundaries with stone was non-existent.
Maraes. The scanty material concerning maraes is presented on page 208. Here again the segregation into three villages prevented the survival of a marae on an outer island.
Graves. (See p. 217.)
Boundary stones (tuakoi) were used in the village of Te Kainga.
Figure 50. Triangular tanged adz (C. 2743). a, front: surface has marked longitudinal convexity; blade and butt surfaces continuous without any distinction except at side edges where shoulders formed by inward inclination of butt; blade slightly convex transversely, defined by well-marked side edges; blade width fairly even throughout; sides curve in slightly to cutting edge, more marked on one side; butt more convex transversely than blade; sides rounded off; adz narrows progressively toward convex poll. b, back: somewhat rounded median edge commences on back of butt, is fairly distinguishable in upper part of blade; in lower part of blade, median edge becomes less distinct and inclines slightly to one side of median line to meet bevel surface; median edge divides back into two postero-lateral surfaces which are convex transversely and meet front at well defined longitudinal edges; front side edges give adz affinity with triangular adz technique and make up for rounded posterior median edge; bevel surface evident, but lower part broken by concave depression which has marred cutting edge; owing to flattening out of median edge, chin curved in outline; upper part of butt slopes upward and forward to meet convex curve of poll. c, right side: front longitudinally convex; back concave; side edge continuous from blade to butt and over poll; butt slopes upward and forward to meet poll edge. d, cross-sections.
1. Tanged adz (fig. 50; pl. 10, C, 2). The adz is made of a gray stone and has been so weathered that the general surface is smooth. It gives the general impression of a triangular tanged adz with a posterior median edge. The posterior median edge is not pronounced owing to the comparative thinness of the blade and the rounding off of the lower part.
2. Small triangular adz, without tang (fig. 51; pl. 10, C, 1). This fragment is regarded as an adz because it was found near the tanged adz and was of similar stone, though incrusted with a deposit of lime. The lower end was unfortunately chipped off some considerable time ago, for the white deposit completely incrusts it. The posterior median edge is well marked though rounded off by the incrustation. The lower end shows a trace of a triangular bevel surface.
3. Large quadrangular adz, with tang (fig. 52; pl. 10, C, 3). Whatever the doubts that may be occasioned by the preceding artifacts, none can exist with regard to the third adz. It is made of gray stone which has been ground throughout. It is characterized by a well-formed tang defined below by a deep convex shoulder and ending above in bilateral lugs at the junction with the poll. The longitudinal edges separating the four surfaces of the blade are well ground, and the two posterior ones are markedly concave. Owing to the pronounced longitudinal concavity of the back, the chin is prominent. The bevel surface is large and with the blade increases in width towards the cutting edge.
Figure 51. Triangular adz, without tang (C. 2744). a, front: surface slightly convex longitudinally and transversely; lower end shows loss of cutting edge by large chip fractured obliquely downward from before back; sides marked by distinct edges; surface narrows from cutting edge to poll formed by convex curve somewhat lower on one side. b, back: posterior median edge extends from just below poll and meets what was evidently triangular bevel surface at its apex; of bevel surface only trace of upper part remains but is sufficient to indicate that it was triangular; median edge divides back with two transversely convex postero-lateral surfaces which meet front in typical acute angled edges of triangular adzes. c, right side: longitudinal front edge continuous between blade and butt and continuous over poll. d, cross-sections.
Figure 52. Quadrangular adz, with tang (C.2742). a, front: edge narrows slightly owing to rounding off of corners; adz slightly concave longitudinally and flat transversely except for slight rounding off of well defined side edges (3); long flaw (4) along one of side edges; shoulder (2) flat and forms convex curve toward tang (5) formed by grinding away front of butt and rounding off front longitudinal edges; tang inclines backward at angle with blade and narrows toward poll, where it ends in two well shaped bilateral lugs (6,6) which come to rounded point with hollow between them. b, back: tang and blade surfaces continuous, concave longitudinally and slightly convex transversely owing to slight rounding off of back longitudinal edges (7); back narrower than front, and parts of lateral surfaces (9) may be seen; bevel (10) forms large surface, quadrilateral in shape, bounded above by straight chin (8), below by wide cutting edge, and at sides by well marked edges formed with lateral surfaces; bevel convex longitudinally and flat transversely, except near cutting edge where slightly concave. c, right side view: shoulder (2) and lugs (6) prominent; back has marked longitudinal concavity; poll (11) forms surface bounded by straight edge at back and concave edge in front due to contour of lugs; cutting edge well ground but blunt and shows no flaws or chips.
The quadrangular adz is exactly similar to an adz (fig. 53) discovered by the late Captain Allen on the island of Nassau, which lies to the west of Rakahanga. This adz was revealed by the slipping away of a bank, when it was seen sticking out from below the surface level. The adz was deposited by Captain Allen in the Australian Museum and afterwards given by him to the Auckland Museum, where it now is. It has been figured by Skinner (23, p. 92), who drew attention to the two lugs (poll-knobs) resembling those in Moriori adzes from Chatham Islands and in a less pronounced form from the South Island of New Zealand. The presence of lugs in the triangular adzes of the Austral Islands has also been remarked by Stokes (manuscript in Bernice P. Bishop Museum). The Nassau adz is a little larger than the Rakahangan adz, as the following figures, compared with measurements made by Skinner, show:page 142
|Length||12 inches||10¼ inches|
|Cutting Edge||3⅜ inches||3 inches|
|Width at poll||1¾ inches||1¼ inches|
|Thickness at bevel||2 inches||2 inches|
|Weight||5¼ pounds||4 pounds (±)|
Figure 53. Quadrangular tanged stone adz with bilateral lugs, from Nassau Island: a, front; b, right side view. Convex shoulder (1) formed by cutting away front of butt to form tang with projecting bilateral lugs (2), marked longitudinal concavity of back (3), clearly defined heel (4) with corresponding thickness of blade and long slope of bevel (5) are also features of Rakahangan adz. (After Skinner.)
Figure 54. Adzes of Tridacna shell. a, adz in B. P. Bishop Museum (C. 2767): shell about 8 mm. thick but thinner in places owing to natural grooving of outer surface; outer surface of shell forms front, ground down in places to remove natural roughness; lower cutting edge fairly straight but curved up slightly at one end; shell fairly straight; bevel to form cutting edge, on front. b, adz in B. P. Bishop Museum (C.2768): outer surface of shell forms front; shell curved, front distinctly convex longitudinally and transversely; lower cutting edge convex; bevel grinding on back; thickness, 5 mm.
Skinner states that the Nassau adz is of light gray volcanic tuff, and it is evident that the Rakahangan adz is of the same material. Little doubt can exist that the two adzes came from the same workshop, being formed of the same material and by the same technique. It would be interesting to know where that workshop was, for both adzes have evidently been introduced, if not by the same people, at least by people who had mutual contact in some part of Polynesia.
Shell Adzes, Haft
Two Tridacna shell adzes (fig. 54) were among the presents given to me on Manihiki. Though they were evidently made for the occasion, it is presumed that the form and hafting reflects the technique of the past. One of the adzes was hafted to a movable socket.
Figure 55. Adz haft, with simple mesial peg. a, side view: rounded branch shaft (1) meets trunk foot (2) with heel (3) above and toe (4) below; upper line of shaft runs straight to heel without forming heel angle, toe angle (5) formed below; toe cut to deeper level at shoulder (6) to form adz surface (7); mesial peg (8) in position; diameter of shaft, 1 inch; depth of foot, 4.6 inches; depth of adz surface, 2.2 inches; foot, 0.7 inch thick at lower end. b, front of foot: transverse shoulder (6) defines upper limits of flat adz surface (7) which is 1.4 inches wide below and 1.5 inches above; foot cut away in mesial line above shoulder to fit mesial peg which is 3.7 inches long and about 0.3 inch thick, and corresponds in depth to shoulder; peg, 0.6 inch wide at bottom, narrows gradually to 0.4 inch at point 9, widens out again to 0.6 inch, gradually diminshes to 0.35 inch, and projects above toe for 1.3 inches; peg thus doubly wedge-shaped to prevent upward displacement from pressure of adz poll when in use.
Figure 56. Lashing of simple haft: shaft, 1′; heel, 2′; shoulder, 4′; median peg, a, 5; adz, d-f, 6′. a, side: one end (1) of lashing braid crossed over middle line on back of heel and braid (2) brought down on near side of shaft, under toe angle, up on far side, and crossed (3) over end in upper middle line to fix it; braid continues horizontally (4) across toe, crosses front obliquely downward, and is brought down obliquely on far side to toe angle where it crosses first turn (2). b, side: from toe angle braid ascends obliquely (5) on near side, crosses obliquely upward over previous turn (4) in middle line in front, passes horizontally backward on far side of heel, and again crosses over previous turn (4) in middle line on back of heel to descend (6) on left of previous turn (2) to toe angle; one set of turns complete. c, front, crossing of descending turn (4) with ascending turn (5) in front middle line, after first set of turns. d, side: two more sets of turns made; thus from b, braid passed up on far side, crossed in middle line above to form second horizontal turn (7) above first one (4), crosses front and far side of foot diagonally downward to toe angle, ascends on near side to make diagonal turn (8) above first one (5), crosses horizontally backward on far side of heel, and makes downward turn (9) on near side to right of first buried turn (2); second set complete; third set ascends on far side to right of previous turns, crosses above in middle line to make horizontal turn (10) above previous turns (4,7), descends obliquely across front and far side to toe angle, makes ascending oblique turn (11) above previous ones (5, 8), crosses horizontally on far side of heel, crosses over middle line above to form last turn (12) to right of previous turns (6,9); repeated chevron pattern complete but continuous braid (12) carried down to make next series for fixing adz to foot; adz (6′) placed in position on foot and series of four loose turns (13–16) made over left thumb laid against toe, first turn (13) lowest and subsequent turns thus inclose descending braid (12); after last turn (16), end (17) of braid passed down through loose loops; loops (13–16) low in figure to avoid confusion with upper lashing, but in practice higher so that upper turn (16) will correspond with level of shoulder (4′). e, front: in repeated chevron pattern in upper lashing, ascending turns (5, 8, 11) cross in middle line with descending turns (4, 7, 10); loose turns of lower lashing have been drawn taut in turn, commencing with lower one (13), finishing with uppermost (16) which is level with shoulder and overlaps it; when last turn drawn taut, slack removed by pulling it down under turns by hauling on lower end (17 in d). f, side: from first transverse series, braid (17) on removal of slack carried down to make another series of four looped turns as in d, commencing from below upward, with end (19) pushed down under them from above; turns (18) drawn taut successively around lower end of foot; slack drawn taut and end (19) cut off close below lowest turn.
Figure 57. Haft with socket, and lashing technique. a, b, side and front views of haft: 1, shaft, 13.2 inches long and slightly more than 1 inch in diameter; 2, foot, 4.5 inches long, front grooved throughout length to fit back of rounded socket; 3, heel, 0.6 inch deep and 1.1 inches wide; 4, toe, 0.8 inch deep and 1.5 inches wide; 5, toe angle; 6, heel angle; differs from simple haft in presence of heel angle. c, d, front and side views of socket of hard ngangie wood: 7, blunt pointed knob, 1.2 inches high, 1.1 inches wide, and 1.1 inches thick at base; 8, heel shoulder, extends from back to sides and rests on haft heel; 9, shoulder, 0.4 inch deep and 1.4 inches wide; 10, flat surface 1.6 inches long cut to shape of adz butt; 11, toe shoulder, section 4.6 inches long between toe shoulder and heel shoulder rounded off to fit against groove of haft foot and allow socket to rotate; 12, slight flange on back, section 0.3 inch long between flange and inward cut below toe shoulder receives adz lashing; length of socket, 7.6 inches; diameter at lower end, 1.3 inches; thickness at base of knob (8), 1.1 inches and just below knob, 0.8 inch; thickness at shoulder (11), 1.4 inches and just above, 1.1 inches; thickness at lower end, 0.6 inch. e, f, lashing, side and front views: heel shoulder of socket fits above haft heel (3) and its toe shoulder (11) below haft toe (4); socket first attached to haft by repeated chevron lashing (14) which makes turns around heel and toe angles, technique of turns similar to that in simple haft lashing (fig. 56, d); second lashing of transverse turns around socket and toe is commenced with slip knot (15) for first upper turn; turns pass from right to left on near side; last three turns made over thumb, end (16) passed upward beneath them, turns drawn taut from above down, slack drawn taut, and end cut off; adz (13) fitted to adz surface on lower end of socket with poll resting against adz shoulder (9) of socket; lashing commenced by turning down one end (17) of braid on adz and making number of loose transverse turns over thumb under which other end (18) of braid turned up from below; turns drawn taut from above down, first three turns fixing and concealing first end (17); last turn (19) drawn taut by pulling on end (18) to remove slack; slack cut off and end fixed; commencement and ending technique differ from that in simple lashing (fig. 56, d); end (20) of braid laid obliquely across middle, braid forms lowest ascending turn on right, and after passing around right side of heel crosses middle line above, descends on left or near side of shaft (e, 20), crosses under toe angle, ascends shaft on far side to cross itself above (e, 21), whence it crosses toe (e, 22) on near side to appear on left side (f, 22) to descend obliquely over front and cross commencing end in middle line; this establishes order of first set of turns and three more sets follow; last ascending turn (23) carried over middle line and tied with overhand knot (24) around two upper turns on right to fix lashing.
1. Simple haft (fig. 55, pl. 10, C, 5). The front surface of the foot is cut away to form a projecting shoulder to prevent the adz from working up on the foot. A slot, in which a mesial peg of hard ngangie wood has been fitted, has been cut in the front middle line. The mesial piece is wedge-shaped with the base below, and the upper end projects above the heel to form an ornamental projection. The inserted piece is of harder wood than the haft and has evidently been used to replace the pith canal of the foot and thus give the upper end of the adz something hard to rest against. The method of lashing the adz (fig. 54, a) to the simple haft is shown in figure 56. The decorative motive used is the repeated chevron which is produced on the front, the back of the heel, and on each side of the toe angle. This lashing, besides being decorative, holds the median peg in position, but does not affect the adz. From this first upper lashing, the braid is carried down and lashes the adz to the toe by two series of transverse turns.
2. Socketed haft (fig. 57, pl. 10, C, 4). The adz with socketed haft was obtained on Manihiki, and the type was said to have been used in the atolls in olden days. A socket has been defined (4, p. 178) as “an intermediate wooden piece by which the adz may be attached to the haft. It may be fixed or rotating.” The Manihikian haft, socket, and lashing of the adz are described in figure 57.
In discussing the socket, Skinner (4, p. 178) states, “In Polynesia the primary function appears to have been the attainment of greater security of attachment, though in the Ellice Islands, rotation of the adze or axe was the principal function.” The rounded character of the Manihikian socket enabled it to be rotated in the rounded groove of the haft. The two upper lashings attached the socket to the haft, but it could be rotated quite readily under the lashings. The third lowest lashing fixed the adz immovably to the flat surface cut on the lower end of the socket. By grasping the lower end of the socket with the adz, the line of the cutting edge may be turned obliquely to its normal transverse position with regard to the long axis of the haft. In this position, it was easier to dub out the narrowing hold of a canoe toward the bow or stern. The craftsman could still swing the implement in the long clear axis of the hold and adjust the socket to suit any angle and either side.