Arts and Crafts of the Cook Islands
Implements with which to fashion wood into houses, canoes, food utensils, weapons, and other objects were of primary necessity. In atoll islands devoid of stone, Tridacna shell furnished the most suitable raw material; but in volcanic islands, the more durable basaltic stone was used. In the Cook Islands, page 134the abundant basalt was worked into adzes and chisels. The making of stone implements reached a high standard of excellence and became a specialized craft, hence the Cook Islands implements were among the best made and the best finished in Polynesia. The following account of adz making, given by the Mangaian pastor, Mamae (29, pp. 117-119), is the only record given by a native who actually saw the process, so I quote it in full.
The best basaltic stones for making axes [adzes] are found at the head of the valley Mataare, or else at Rupetau. If a man wanted some axes made he would go to one of these places, and with his ironwood spade dig out the best stones he could find, and carry them to the artisan who had agreed to make a set of axes. Payment was rendered in food and cloth. The taunga (=artisan) takes up a large piece of red flint, and begins vigorously to chip away at one of the stones. After working at it for several hours, and getting the rough outline of the axe, alas! the stone breaks off in the centre! Time and labour have been wasted. Taking up the broken fragments, he says, "ti-ria-tu!" or "throw it away!" (expressive of disgust).
He now sets to work on a second piece of basalt. To his annoyance, much larger bit's than he intended fly off. He chips with greater care and gentleness than before; but again and again pieces fly off. The stone does not break in two, as in the former instance; but the size of the axe must be reduced, its value being proportionally lessened. He lays aside the heavy piece of flint and uses instead a much lighter one, hoping eventually to be rewarded for his trouble. But again large pieces fly off as before, proving the stone to be essentially worthless. This, too, is thrown away.
"Ugh!" says the axe-maker; "maybe the next will turn out no better. My labour is in vain!" He this time carefully scrutinizes the quality of the stone he intends to operate upon. The red mountain clay, which conceals its real colour and defects, is washed off. As the stone seems to be really excellent, the man cheerfully sets to work upon it. Chip—chip—chip goes the large sharp flint. The stone yields, and under the patient hand of the taunga soon assumes the form and appearance of a first-class axe. A lighter piece of sharp flint is used for the finishing off. Now only fine dust falls, blown away from time to time by the breath of the much-enduring workman. The axe is almost ready for polishing, when the taunga detects an inequality in the surface; in the correction of this fault, alas! the stone breaks in two. On carefully examining the fragments, there is seen a flaw in the centre, though externally all was faultless.
Is this laborious workman to reap no fruit of his toil? Will all these stones, dug out of the sides of the distant mountain, prove worthless? Not all, it is hoped.
Again he goes to work upon a piece of stone—a really good piece of basalt. Chip—chip—chip goes the big flint. After a time a lighter flint, equally sharp, however, is used. The work proceeds slowly but prosperously. Whilst an ignorant observer wonders why the shapeless stone is battered, bruised, and unceasingly chipped, the master knows that every blow is needed, and tends to the perfection of his work. The black stone under the small bit of red flint becomes covered with fine white dust, requiring to be frequently removed. At last the polishing alone remains. Pieces of coral, broken off the reef, are rubbed long and laboriously upon the axe, often wetted with water. Eventually the axe is completed, to the satisfaction of the maker and to the envy of others.
Throughout the above passages substitute adzes for "axes" and ta'unga for "taunga." The red flint for chipping the basalt was termed ruarangi and it is evidently the chert form of silica referred to by Marshall (52, p. 37), who says, "On the slopes above Ivirua huge boulders of chert strew the hillsides. The more solid boulders yield chips with the typical sharp edges. …"
The hammer stones of other islands are usually spherical, or at least without an edge. Mamae's statement that the large hammer stone was sharp may page 135apply to the chert flakes of Mangaia. It is also interesting that a smaller flake was used for finishing off. The processes used were chipping and then pecking and bruising as indicated by the statement that in finishing off, fine white dust had to be blown off. The coral used in the final rubbing down was the form termed aravai. I was told that tongaitu, which grew on submerged coral masses, was dried in the sun and used as a kind of sandpaper to rub the adzes.
No mention is made in the above description of sharpening the cutting edge. The process of rubbing the adz head back and forth on a block of stone to sharpen the cutting edge was termed oro, but I did not see any grinding stones.
After contact with early European voyagers the Cook Islanders, like other stone-age people, discovered that metal tools were vastly superior to their own stone tools. Nails, hoop iron, and steel hatchets were the most desirable objects that the foreign culture could provide. Traders and missionaries continued the supply of foreign implements, and metal tools gradually replaced stone tools. Thus the craft of making stone implements ceased over a century ago, long before other crafts died out. Though they are still found in the various islands, the present population can give no authentic information as to the exact uses of the various types. However, as pieces of hoop iron or plane blades were lashed with sennit to handles made on the old pattern, the old type of handle and the lashing technique continued to function until fairly recent times.
The stone tool in general use throughout Polynesia was the adz, beveled on one side to form the cutting edge and hafted so that the cutting edge was at right angles to the long axis of the handle. In Mangareva, however, axes that were equally beveled on two surfaces were used in addition to adzes (77, pp. 259, 269). A few stone implements have been found in other islands with two bevels at the cutting edge, but there is no way of knowing how they were hafted. In dubbing out narrow spaces such as the bow and stern ends of a canoe an ax would have been a better tool than an adz. Polynesians in general seem to have used the adz exclusively, but in some island groups the difficulty of working in narrow spaces was met by attaching the adz-head to an intermediate wooden socket which in turn was attached to the haft in such a way that the socket could be twisted to turn the cutting edge of the adz into a longitudinal axis to perform the functions of an ax. I have not seen a socketed adz from the Cook Islands, but the Mangaians say that such adzes, termed toki tupa, were used for working in the bow and stern of canoe hulls and in small canoes in which the sides were undercut from the top edge and in the wooden gongs (ka'ara) that had to be hollowed out from a narrow slit on the top side.
The Hawaiian reversible or socketed adz was termed kupa-ai k'e, in which kupa is the same word as the Mangaian tupa, with the Hawaiian dialectical change of t to k. Malo (50, p. 171) said that the inside of the canoe was finished page 136off with the kupa-ai ke'e, and the Mangaians stated, "E toki ta ma te toki tupa" (An adz for finishing off was the tupa adz). The socketed adz was known from Hawaii to New Zealand, and the use of the term tupa in Mangaia and Hawaii (kupa) shows that this term must have been in vogue in some common center before diffusion took place.
I studied Cook Islands adzes on specimens from Aitutaki, Mangaia, and Rarotonga in the Auckland Museum and on specimens in a private collection in Rarotonga (70, pp. 213-244). This study is based on material in Bishop Museum (table 1) which includes adz heads from Atiu and Mauke. The total series is too small for mathematical deductions, but it is sufficient to give general indication of the prevalence of certain features in Rarotonga, Atiu, and Mauke. The collection consists of medium-sized and small adzes and one large specimen. Broken adzes or fragments that clearly illustrated some feature of construction were included in the study when their locality was known.
|Island||Inverted Triangular Adz||Quadrangular Adz||Total|
The material was classified as, triangular or quadrangular according to a section through the blade above the posterior bevel surface. The normal triangular form of adz has the base or flat surface to the back. Though this type, with variations, is widely spread throughout Polynesia, it does not occur in the, series under discussion. An unfinished Rarotongan adz of this type was figured in a previous study (70, p. 232). The triangular form in this series has the base or flat surface to the front and is termed the inverted triangular form, but in the following discussion, the term triangular refers to this form. The terms used are defined in the Polynesian Journal (vol. 14, pp. 174-180) except that "adz head" may be used instead of "adz" to distinguish the stone implement without its haft (77, p. 261).
Inverted Triangular Adz Heads
The inverted triangular type of adz (table 1) is the commonest form in the Cook Islands. The flat surface, or base of the triangular section, is to the front; the back is formed by two postero-lateral surfaces which meet in a posterior median edge. A number of differences exist in the treatment of page 137various parts of these adzes, but I regard them as variations from one general type and will not attempt to classify them into types and subtypes as Stokes did in his exhaustive study of the adzes of Tubuai (Aitken, 1, pp. 130-155).
The adz head is divided into an upper butt that is lashed to the haft and a lower blade which extends below the lashing. The butt is usually much shorter than the blade, but in some the length of the butt may approach that of the blade or, in rare instances, may even exceed it. In a Rarotongan fragment of a fairly large adz, the butt is so short that it is difficult to understand how there could have been sufficient space for a lashing to hold it in position. In most of the implements, the butt is divided from the blade by a distinct shoulder, but there may be rare ones with no shoulder.
It may be assumed that the front surfaces of the butt and the blade were in the same plane when an adz was chipped into shape. In one type of quadrangular adz (fig. 93) there was no technical indication of the junction between butt and blade; it was the craftsman who decided how much of the adz should serve as butt and be included under the lashing to the haft. The part below the lashing became the blade. The junction between butt and blade in such adzes was thus arbitrary and was only apparent in lashed adzes. The removal of the haft also removed the arbitrary junction established by the craftsman.
Figure 73.—Triangular adz, without shoulder, Mauke (B8709). a, front: wider at cutting edge than poll, clearly defined lateral edges; b, back: high bevel apex and posterior median edge extending to poll; c, right side: no trace of working lateral edges of butt; d, sections. Weight, 5.25 oz.
Triangular adzes are characterized by a distinct shoulder, which definitely indicates the junction between butt and blade. In this series, one specimen is an exception in that it has no trace of a shoulder (fig. 73); the front surface is continuous without interruption, and the sharp lateral edges of the blade are continuous with sharp lateral edges on the butt. This adz may be the page 138survivor of an early form, or, as is more likely, be the product of inferior craftsmanship.
Figure 74.—Triangular adz, with slight lateral shoulders, Mauke (C2711). a, front narrower at cutting edge, slight shoulders formed at sides by rounding off lateral edges of butt; b, back: bevel surface concave transversely and median edge continued to poll; c, right side: sharp lateral edges of blade contrasts with rounded edges of butt. d, sections. Weight, 5.75 oz.
The characteristic shoulders were formed on the front and were produced by two techniques which are opposite in principle. One technique consisted of leaving a forward projecting ridge, the other of chipping down the front surface of the butt.
Figure 75.—Triangular adz, with raised shoulder, Cook Islands (B3522). a, front: increase in width from poll to cutting edge, raised shoulder between blade and butt which are on same plane; b, back: concave bevel surface and median edge extending to poll; c, right side, raised shoulder evident in profile, side edges of butt rounded off;d, sections. Weight, 9.75 oz.
In most adzes, the butt shoulder is formed without leaving a raised ridge, but the rounding off of the butt lateral edges is extended to include the whole of the front surface of the butt. With the adz held horizontally in the position of manufacture, the front surface of the butt drops to a lower level than that of the blade. This type of shoulder may therefore be termed a "dropped shoulder" in contradistinction to the raised shoulder. The shoulder forms a distinct edge with the upper boundary of the blade and varies in depth according to the amount of the butt front surface removed. It is usually deeper in thick adzes than in thin adzes. The transverse line of the dropped shoulder may be straight or slightly curved downward at the outer ends; or in some, it may be convex upward. What would have been lateral shoulders become merely the ends of the dropped shoulders (fig. 76).
Figure 76.—Triangular adz, variation 1, Atiu (C2721). a, front: long, comparatively narrow blade and poll, narrower cutting edge and clearly defined lateral edges to blade, dropped shoulder formed by working down front surface of butt, lateral edges of butt rounded off; b, back: short bevel surface, straight transversely; median edge continued to hollow at poll; c, right side: sides of butt and blade markedly convex transversely, profile shows dropped shoulder with butt front surface on different plane; d, sections. Weight, 28.25 oz.
Figure 77.—Triangular adz, variation 1, Mauke (B8711). a, front: comparatively narrow butt and blade with narrower cutting edge, raised-dropped shoulder; b, long bevel surface, straight transversely, and bevel apex on butt; median edge continued to poll; c, right side: profile shows up raised-dropped shoulder; d, sections. Weight, 20.25 oz.
Of the series examined, 35 percent had the raised shoulder and the raised-dropped shoulder with more of the latter. The percentage varied in individual page 141islands, however. Though the numbers are small, the percentages of each island series may give a general idea of the relative occurrence of the forms. Atiu had 55 percent, Mauke 46 percent, and Rarotonga 17.5 percent. Of the four adzes from Aitutaki, two of them had raised-dropped shoulders. Mangaia had two with the pseudo raised-dropped shoulder. It may be said that the raised and the raised-dropped shoulder adzes are common in the three northern islands of the group?Atiu, Mauke, and Aitutaki; that they are fewer in Rarotonga; and that they are absent from Mangaia.
A rare form of shoulder occurs when the butt is given a backward incline so that the surfaces of the butt and the blade meet at an obtuse angle, which forms a transverse line at their junction in front. This form is common in Hawaii and may be termed an "angled shoulder." It occurs in a thin quadrangular adz from Rarotonga (fig. 95).
The front surface of the blade is bounded by the shoulder above, the cutting edge below, and lateral edges on either side. The lateral or front longitudinal edges are sharply defined, as they form an acute angle with the postero-lateral surfaces which slope inward and backward toward the middle line. The thinner the adze, the sharper the edges. The front surface may appear flat but it is convex transversely; and, viewed from the side, a part of the front surface will always show beyond the lateral edges. The transverse convexity is particularly marked in adzes which have a concave beveled surface to the cutting edge. Longitudinally the surface is also convex, more sharply so toward the lower end near the cutting edge.
The front varies greatly as to relationship between length and width, some being long and narrow, others short and wide. Some have approximately parallel sides, others are narrower at the edge than at the shoulder, and others are wider at the edge. In the series under discussion, 51.7 percent are narrower at the cutting edge, 14.3 percent are equal, and 34 percent are wider.
The back of the blade is formed by what I have termed the two postero-lateral surfaces meeting in a posterior median edge. The lower part is chipped and ground to form a bevel that meets the lower edge of the front surface to from the cutting edge. Because the cross-section of the adz is triangular, the beveled surface leading to the cutting edge is triangular with the base on the cutting edge and the apex on the posterior median edge. The bevel apex varies in position with regard to the top of the blade defined by the front Shoulder. In thick adzes, this apical point is situated at from a third to two thirds of the length of the blade. The apical point in thinner adzes reaches to the top of the blade, and in the series under discussion this is true in 44 percent. In the Mangaian adzes, however, the percentage is much higher. Of page 142the 11 Mangaian adzes in this series, 8 (73 percent) have the apex at the top of the blade. A series of 27 described by Dodge (21, pi. IX) has 81 percent. Though the majority of the adzes are of the thin type, the data, Indicate that the Mangaian adzes had a longer bevel surface than those of the other islands.
In the Cook Islands adzes the bevel surface is commonly concave transversely, and the front surface is more convex transversely than in adzes with a straight bevel surface. The concave bevel surface is present in 44 percent of the adzes from all islands, but individual islands vary as follows: Atiu, 60 percent; Mauke, 50 percent; Rarotonga, 36 percent. Of the small series of 4 from Aitutaki, all are concave. Of 11 from Mangaia, only one is slightly concave. The cutting edges of such adzes are curved with the convexity forward. In use, they must have made grooves like those made by a gouge.
When the bevel apex does not reach the level of the shoulder, a median edge is formed by the two postero-lateral surfaces meeting above this apical point. In thick adzes, the median edge is sharply defined; but in thin or very wide adzes, it may be rounded off. In some adzes that are comparatively thin in proportion to width, the two postero-lateral surfaces may not meet in the middle line; consequently, a narrow posterior surface is formed which is continued on the butt (fig. 80). Thus, a section through the blade above the apical point is quadrangular. In a previous study (70, p. 233), I regarded adzes with a little wider posterior surface as quadrangular adzes, but the slope of the sides and the treatment of the butt in forming a shoulder leads me now to regard such adzes as a variation of the inverted triangular technique due to the nature of the piece of stone used.
The lateral edges of the blade are sharply defined throughout their length from butt shoulder to cutting edge. In thin adzes or adzes with wide blades, the edges are sharp, being formed by an acute angle. In thick blades the edges are more obtuse. In many of the Mangaian adzes, the sharp edges are ground down to a narrow lateral surface which decreases in width toward the cutting edge. This lateral grinding occurs particularly in ceremonial adzes, which are marked by their comparative thinness. In the collection of the Peabody Museum, Salem (studied by Dodge, 21, pl. IX), the narrow lateral surface is found in 16 (59 percent) of 27 edges. In a series of 15 adzes attached to working handles, examined by me in various museums, the lateral surface is found in five (33.3 percent). As the grinding down of the lateral sharp edges into narrow surfaces does not occur in any of the adzes from the other islands of the Cook group in the Bishop Museum series, we may regard it as peculiar to Mangaia and of some importance in the identification of adzes not originally localized. The technique was probably first used on thin adzes to blunt the. sharp lateral edges and later applied to other adzes as a finishing touch; (fig. 83).
The butt follows the triangular section of the blade but diminishes slightly in width and thickness toward the poll. The postero-lateral surfaces continue in the same plane as those of the blade. In most adzes, the posterior median edge of the blade continues, without break, to the poll. In some adzes, particularly thin ones, the median edge is rounded off on the butt. In those that have a narrow posterior surface because the postero-lateral surfaces do not meet, the narrow surface on the blade is continued to the butt. In the Museum series, the prevalence of the three types is as follows: sharp median edge, 63 percent; rounded edge, 28.5 percent; narrow posterior surface, 8.5 percent. However, there is considerable variation in the different islands. In Mangaia and Rarotonga, the posterior median edge is found in over 80 percent; in Mauke, 52 percent; and in Atiu, 35 percent. The Aitutaki series of 4 had one with the median edge.
The rounding off of the lateral edges of the butt and the working down of the upper surface are discussed in the description of the shoulder (p. 139). In butts with a posterior median edge, the section is triangular with the basal angles rounded off; in those with the posterior median edge rounded off, the section is ovoid.
In most adzes, the butt and blade are in one straight axis. In others, the butt may appear to have a backward incline owing to deeper grinding towards the poll end; but, if the back of the adz is examined, butt and blade are seen to be straight (fig. 82). In rare quadrangular adzes there is an actual backward slant of the butt with a resultant angled shoulder (fig. 95). A number of Mangaian adzes have the backward slant; but, owing to the working down of the butt front surface, the shoulder is of the dropped form (fig. 83).
When reduced, the butt forms a definite tang which is characteristic of the inverted triangular form of implement. In thick adzes with a raised shoulder, the rounding of the butt lateral edges hardly constitutes a tang, but when the shoulder is of the dropped form, the formation of a tang is more obvious. In wide adzes, the butt has to be reduced considerably at the sides to a width that is convenient for fitting to the haft; and in such adzes, there is a pronounced tang. The working down of the butt lateral edges may be carried down the sides for some distance with the result that the shoulder is carried down laterally as well. In some Mangaian adzes, the lateral extensions of the dropped shoulder may extend down to the posterior median edge (fig. 83).
Variations in triangular adzes
Adzes were made in various sizes and shapes for different uses; but, lacking authentic information about these functions, I have arbitrarily divided the inverted triangular adzes into two main groups: (1) narrow thick adzes and page 144(2) wide thin adzes. Between typical specimens of the two groups range a great many intermediates. Among the smaller adzes are aberrant forms which partly conform to the general pattern, probably owing to the nature of the flake before it was worked.
Variation 1. Narrow, thick adzes. The blade is comparatively narrow in proportion to length, but the outstanding feature is the fact that at the shoulder the thickness of the blade is nearly equal to or greater than its width. Owing to the comparative narrowness of the blade, little material has to be removed from the sides of the butt in order to fit it to the haft. Consequently, the lateral extension of the shoulder is slight. The proportion between width and thickness can be expressed by the following width-thickness index:
In adzes with a raised shoulder, the measurements should be taken just below the shoulder. In variation 1, the index will range from 90 to over 100. An Atiu adz that is thicker than it is wide, with a width-thickness index of 106, illustrates this group (fig. 76). A Mauke adz of equal width and thickness thus gives a width-thickness index of 100 (fig. 77). These two adzes are of medium size and heavy. A small, lighter adz from Rarotonga conforms to the group test, being thicker than wide and returning an index of 105. It is also one of the few adzes in which the butt is longer than the blade (fig. 78).
Figure 78.—Triangular adz, variation 1, Rarotonga (C2753). a, front: long butt, dropped shoulder; b, back: bevel surface straight, median edge to poll; c, right side; d, sections. Weight, 8.25 oz.
Figure 79.—Triangular adz, variation 2, Atiu (C2731). a, front: badly chipped on left of cutting edge, raised shoulder; b, back: concave bevel surface with high apical point, butt rounded; c, right side: note raised shoulder with planes of butt and blade on same plane; d, sections. Weight, 21.75 oz.
Figure 80.—Triangular adz, variation 2, Atiu (C2730). a, front: raised-dropped shoulder; butt lateral edges somewhat sharp; b, back: bevel surface concave, no apical Point as postero-lateral sides do not meet and narrow surface between continued to poll; c, right side: shows up raised-dropped shoulder; d, sections. Weight, 13.75 oz.
There were no typical forms of variation 2 among the Rarotongan series.page 146
Figure 81.—Triangular adz, variation 2, Mauke (C2713); highly polished on butt and blade. a, front: raised-dropped shoulder, sides of butt well rounded and polished; b, bevel surface concave with high apical point and median edge rounded off on butt; c, right side; d, sections. Weight, 10.75 oz.
Figure 82.—Triangular adz, variation 2, Mangaia (B3635) black basalt, a, front: long blade ground down below shoulder giving false appearance of raised-dropped shoulder but really dropped shoulder form; lateral edges of butt ground in such way as to lower lateral edges below original plane, b, back: concave bevel surface with high apical point and median edge extending to poll. c, right side: showing back slope of butt, also lateral edges of blade slightly touched to blunt sharp edge, d, sections. Weight, 22.25 oz.
Figure 83.—Triangular edge, ceremonial of variation 2, Mangaia (7860); very dark basalt, with well polished blade but unpolished butt. a, front: short butt with somewhat angular lateral edges; blade long, and wider at cutting edge which is convex downward; dropped shoulder. b, back: long bevel surface with apical point at butt junction, middle part slightly concave transversely but straight near cutting edge; dropped shoulder on front continued on sides and back; butt badly chipped on one side, median edge present. c, right side: butt slants backward and lateral edges of blade ground down to narrow surface. d, sections. Weight, 32.75 oz.
A large ceremonial adz in the British Museum is peculiar in having raised ridges on the front and back (fig. 84). Giglioli (27, p. 300) refers to two other adzes with notched ridges; one in the Archaeological Museum at St. Germain near Paris and the other in his own collection (no. 1493). The making of such adzes with symmetrically curved ridges entailed the greatest skill.
Figure 85.—Triangular adz, intermediate, Rarotonga (C8848). a, front: long blade with practically parallel sides an deep dropped shoulder, butt with nicely rounded lateral edges and squared poll end; b, back: bevel surface concave with apical point fairly high and median edge continued to poll; c, right side: shows deep dropped shoulder and extension of shoulder down sides; d, sections. Weight, 65 oz.
Figure 86.—Triangular adz, intermediate, Rarotonga (C2390). a, front: blade polished, butt with pecked surface and well-rounded lateral edges, dropped shoulder with lateral extensions; b, back: bevel surface straight transversely and markedly convex longitudinally; median edge extends to backward slant of poll surface, c, right side: shows depth of dropped shoulder and extension down on side, d, sections. Weight, 8.25 oz.
Figure 87.—Triangular adz, intermediate, Aitutaki (C2760). a, front: butt and blade in same plane but separated by a raised shoulder; b, back: bevel surface concave with apical point above level of shoulder and median ridge extending to poll; c, right side: showing plane of butt and shoulder on same level with raised shoulder between. d, sections. Weight, 12.5 oz.
Figure 88.—Triangular adz, intermediate, Mauke (C2719). a, front: butt and blade on same plane with pronounced raised shoulder; b, back: narrow bevel surface, straight transversely and apex lost in rounding off of blade and butt; c, right side: shows pronounced profile of raised shoulder; d, sections. Weight, 9 oz.
Figure 89.—Triangular adz, intermediate, Mangaia (C1451). a, front: blade smoothed, dropped shoulder with gradual slope to butt surface, butt lateral edges distinct; b, back: bevel surface straight and bevel apex at about half length of blade; median edge extends to poll. c, right side: showing comparative thinness of adz and slope of dropped shoulder. d, sections. Weight, 10.5 oz.
Figure 90.—Triangular adz, intermediate, Mangaia (private collection). a, front; finely polished butt and blade, well-marked dropped shoulder, sharp butt lateral edges; b, back: short, straight bevel surface but upper part slopes away in rounded median edge which continues to poll; c, right side: shows dropped shoulder with extension down sides to posterior median edge. d, sections. Weight, 1.25 oz.
What I have termed variations 1 and 2 are pronounced forms at two ends of the series of inverted triangular adzes. Between these extremes range a large number of intermediate forms with leanings toward either end of the series. Without knowing their specific uses, a classification into types and subtypes would be purely artificial.
As T have stated, aberrant forms were largely influenced by the original shape of the piece of stone used. Some were so thin that neither the triangular nor the quadrangular pattern could be carried out. Some may have been made by indifferent craftsmen who were content to put an edge to a piece of stone and make some attempt to trim the butt for fitting to the haft. Though the small Mauke adz in figure 91 is too thin to have been trimmed into any recognized type, the craftsman trimmed the butt into a tang with marked lateral shoulders, and a certain amount of grinding on the back shows that within the range of the material the craftsman was influenced by the form of variation 2.
Figure 91.—Aberrant form, Mauke (C2702). a, front: butt and blade in one plane, well-marked lateral shoulders, b, back: sides ground sharply inward forming distinct lateral edges to blade and indistinct lines where sides meet bevel surface; blade and butt flat owing to thinness of adz; bevel surface slopes into back surface without apical point or chin. c, right side: sharp lateral edge of blade and rounded edge of butt. d, sections. Weight, 2.25 oz.
In the collection of quadrangular adzes in Bishop Museum are 11 whole specimens and one important fragment. They are distributed as shown in table 1. They may be readily divided into three distinct types.
Quadrangular adzes with the front wider than the back and without a tang are in type 1. The four surfaces are defined by distinct longitudinal edges. The bevel surface to form the cutting edge is quadrangular with a distinct page 152chin between the bevel surface and the back surface. In the type specimen from Rarotonga, the front edges of the butt are slightly rounded off but not enough to form a distinct tang (fig. 92). In another Rarotongan adz (fig. 93)
Figure 92.—Quadrangular adz, type 1; Rarotonga (C2754). a, front surface: straight transversely and convex longitudinally; gradual increase in width from poll to cutting edge; slight rounding off of front edges of butt by pecking. b, back: shows narrower back surface and shortened view of sides, also bevel surface and chin. c, right side: shows increasing thickness from poll to chin. d, cross sections with front uppermost. Weight, 7.75 oz.
Figure 93.—Quadrangular adz, type 1; Rarotonga (C518). a, front surface: straight transversely, convex longitudinally; front edges of butt continuous with those of blade and not rounded off; increase in width from poll to cutting edge. b, back: shows narrow back surface and shortened view of sides; quadrangular bevel surface with chin. c, right side: increasing thickness to bevel of chin. d, sections. Weight, 3.5 oz.
Some small, thin adzes with a flat back and the butt unworked, fall into type 1, but the sides are somewhat irregular. There are two specimens in the collection, one from Rarotonga and one from Mangaia (fig. 94).page 153
Figure 94.—Small, thin quadrangular adz, type 1; Mangaia (C2748). a, front surface: straight transversely, convex longitudinally; width increases to cutting edge; front edges not clearly defined because chipped hollows are not ground out. b, back: shows rough chipping of sides slanting forward to make back surface narrower than front; bevel surface quadrangular but sloping onto back, not having clear chin line. c, right side. d, sections. Weight, 2.25 oz.
Quadrangular adzes with little difference between the width of the front and back but with a tang are in type 2.
Figure 95.—Quadrangular adz, type 2; Rarotonga (C8961). a, front surface: shows shoulder with slight downward curve at sides; blade convex transversely and longitudinally; increase in width downward; front side edges well marked on blade, rounded off on butt. b, back with well-marked side edges on blade and butt; bevel surface concave transversely with gradual slope onto back, without distinct chin; cutting edge distinctly curved with convexity forward; back surface slightly wider than front. c, right side: shows the backward slant of butt making back concave longitudinally from poll to bevel surface. d, sections. Weight, 4.75 oz.
Figure 96.—Quadrangular adz, type 2; Rarotonga (C8960). a, front: shows short blade and curved shoulder with convexity upward; well-defined front edges; front butt surface and sides ground down to form tang. b, back: back edges clearly defined on blade and butt; back straight transversely and concave longitudinally; quadrangular bevel surface and straight chin. c, right side: shows thickness of adz, shoulder, concave back longitudinally, and angle of chin. d, cross sections. Weight, 5.75 oz.
Figure 97.—Quadrangular adz, type 2 with nipples; Rarotonga (C8962). a, front: shows shoulder and two nipples; front edges of blade well defined and front edges of butt well rounded off. b, back: back edges well defined on blade and butt. c, right side; showing deep shoulder and forward projection of nipples. d, section. Weight, 3.75 oz.
The fact that there are two nipples near the poll end of the Rarotongan quadrangular adz (fig. 97) is interesting, as a similar technique is reported on an Aitutaki adz by Skinner (correspondence). A quadrangular adz with two nipples near the poll was found in Rakahanga (75, p. 141), and a similar adz from Nassau was figured by Skinner (60, p. 92) and Hiroa (75, p. 142). page 155 Skinner drew attention to the two nipples (poll-knobs) resembling those in Moriori adzes from Chatham Islands and in a less pronounced form from the South Island of New Zealand. A quadrangular adz with two poll nipples was obtained in Uvea by Burrows (18, pp. 46, 47), and the Bishop Museum has two such adzes that were collected on Rapa. The wide distribution of this feature may indicate that it was an early technique associated with quadrangular adzes, but the Oldman adz (fig. 113) with two well-marked nipples is of the inverted triangular form. Another triangular adz in the Bishop Museum (C4782) was attributed to Rapa by the dealer from whom it was bought; but, as triangular adzes are not characteristic of Rapa, it is more likely to have come from the Austral Islands where the triangular form is the common type. The whole subject needs further study.
Quadrangular adzes with the front narrower than the back, the front of the butt ground down to form a tang, and the back of the adz concave longitudinally are in type 3.
Figure 98.—Quadrangular adz, type 3; Atiu (C2740). a, front: anterior surface of blade defined above by curved shoulder with convexity upward, and gradually widening to cutting edge; side edges of blade well defined; lateral surfaces seen on either side of blade. Butt slants backward and front surface worked to lower plane with side edges rounded off. b, back: occupied entirely by posterior surface of blade and butt with posterior side edges clearly defined on butt and blade; bevel surface concave transversely forming curved chin with convexity upward; cutting edge curved. c, right side: lateral surface of blade and butt are continuous and slope forward from wide back to narrower front; shoulder is prominent, and longitudinal concavity of back is evident. d, sections. Weight, 7.25 oz.
Figure 99.—Quadrangular adz, type 3; Rarotonga (C8871). a, front: narrower front surface well defined by side edges on lower two-thirds of blade, upper third and curved shoulder somewhat rounded off; butt with front surface and side edges rounded off in concave longitudinal curve leaving a single rounded nipple in middle line near poll; cutting edge chipped on left side and thus breaking normal curved line of original cutting edge. b, back: posterior surface of blade and butt occupies full width of back with side edges well defined; bevel surface slopes into back without a defined chin and toward cutting edge, surface is slightly concave transversely. c, right side: shows poll nipple and shoulder on front and long concave curve formed by butt and blade which changes to convex with forward grinding of bevel surface. d, sections. Weight, 22.5 oz.
Figure 100.—Quadrangular adz, type 3; Mauke (C2707). a, front: front surface of blade narrower than back; lateral edges well defined; dropped shoulder with convexity upward; butt well rounded at sides. b, back: bevel surface slightly concave and sloping into flat back without a chin; back edges of blade well defined but continuation onto chin not so distinct. c, right side: profile of dropped shoulder with extension down sides. d, sections. Weight, 6.25 oz.
In other island groups, the adz butt is usually left rough to give a better hold to the lashing turns; but in the Cook Islands, care equal to that devoted to eliminating chipped hollows and smoothing the blade is used in finishing the butt. In a few adzes with polished blades, the polishing extends to the butt and includes back and front. The usual signs of pecking on the rounded lateral edges are eliminated by the polishing. In some adzes, the poll surface is trimmed so that it forms a perfect section of the butt. In some, however, there are flaws that could not be remedied.
In any large collection of adzes, there are poorly made specimens. Such adzes were probably made by less skilled craftsmen who could not afford to hire the services of an expert. The well-made adzes, particularly of Mangaia, indicate that the craft reached a high standard and that experts developed an aesthetic sense in the trimming of the butts, and so created works of art.
Adz hafts were made from a tree branch which formed the shaft, a block of the tree trunk forming the part to which the adz head was attached. Sometimes a secondary branch was selected and the block taken from the larger branch from which the secondary branch sprung. Sometimes a branch with a marked curve provided both shaft and block.
The haft is divided into the shaft formed from the branch and the foot formed from the tree block (14, p. 176, fig. 2; p. 179). The completed haft is inverted, and in this position the acute angle between branch and block which was above on the tree is now below in the haft. The upper end of the foot is termed the heel, the lower end, the toe. In different parts of Polynesia, the shaping of the foot varies according to the importance given to the toe or the heel. The form of haft is influenced by the form of the adz head, and the method of lashing the adz head is, in turn, influenced by the form of the foot. The hafts used in Polynesia may accordingly be divided into three general types determined by the functional importance of the toe, the heel, or both. For the purposes of comparative study, it is convenient to term the middle part page 158of the block opposite the shaft junction the body, while the projection below forms the toe and the projection above forms the heel. The three general types may be termed the toe haft, the medium haft, and the heel haft. (See figure 101.)
Figure 101.—Types of adz hafts: a, selected branch (1) for shaft, scarfed to provide block (2). b, block removed and inverted so that acute angle (5) is below; middle part (2) opposite shaft junction (1) is front proper, while projection below shaft provides toe (3) and projection above shaft provides heel (4), acute angle between shaft and toe is toe angle (5) and obtuse angle between shaft and heel is heel angle (6). c, toe-haft type, upper projection is trimmed off so there is no heel angle and consequently no actual heel except for heel point (4) where two planes meet; toe (3) is left sufficiently long to support full length of butt of adz head (7) which is fixed by simple transverse turns around toe. d, medium-haft type, both toe (3) and heel (4) are present with their accompanying angles, toe is reduced in length so that adz butt rests on both toe and body and lashing turns consequently pass around toe, body, and heel. e, heel-haft type, heel (4) is prominent with a sharply defined heel angle (6) but there is no toe and consequently no toe angle, lashing turns pass around body, heel, and shaft.
The heel haft was used in all the Cook Islands in the working adzes, and in the highly specialized ceremonial adzes of Mangaia the foot had no toe. The shape of the foot varied somewhat; but, as the foot shape affects the lashing technique, it is preferable to discuss the variations in conjunction with the lashing. In some Mangaian working adzes, the proximal end of the shaft is sometimes flared; in others there is a dorsal edge along the whole length of the shaft.
Sennit was always used in the Cook Islands for lashing. In the medium-sized braid (3.5 to 5 mm. in width) used for working adzes, the coir fibers of the outer ply were twisted over as the ply was turned into the middle position thus rendering the ply narrower and thicker. In the sennit used for the ceremonial adzes of Mangaia, where neatness of pattern rather than firmness of lashing was desired, the plies consisted of a small number of fibers which were kept flat as the outer ply was turned into the middle position. This flat braid termed rapa averaged 2 mm. in width, and patterns obtained with it were closer and neater than was possible with the thicker braid. Sometimes a flat four-ply braid was used instead of the usual three-ply sennit.
The study of lashing patterns lacks representative material that was actually prepared by craftsmen who had used stone tools. Old working adzes in museums can be detected by the deep color of the sennit and the smooth appearance of the shafts through actual Use. Later, adz heads, which were readily picked up, were hafted by natives for trade or for gifts. As the craftsmen of the metal age had no models to copy, it is sometimes doubtful whether the lashing patterns conformed to an original type or were modern makeshifts. In many of these later adzes the bevel of the cutting edge is faced to the front, whereas in all the old adzes in museums the bevel is invariably to the back. It is fortunate for the Tahitian area that so many old adzes with lashings complete had been collected in early times and preserved in museums and private collections. With the exception of Mangaia, the Cook Islands have not been so fortunate. A single adz in the British Museum attributed to Rarotonga bears the signs of age, and the lashing pattern is identical with the Tahitian form. I have seen three adzes from Aitutaki which, though of later date, conform to the Rarotongan pattern. This pattern, which may be termed the "triple triangle" may be regarded as an old form for Rarotonga and Aitutaki; and in all probability it was used in Atiu, Mauke, and Mitiaro (pl. 9, A). The Mangaians were regarded in the Cook Islands as the best workers with sennit, and they have justified their reputation by developing two complicated forms of lashing which were peculiar to themselves. One Mangaian pattern is preserved on a large number of ceremonial adzes and also on a number of working adzes. It is an elaborate variation of the triple triangle and may be termed the "Mangaian triple triangle" (pl. 9, B). The Mangaians used another complicated pattern which may be termed the "multiple lozenge" lashing (pl. 9, C). In the Cook Islands attention was directed to the utilitarian object of attaching the adz head to the haft and to producing an artistic pattern as well.
The triple triangle technique enlarges the three triangles formed by the sennit crossings as the lashing turns continue so that their apices converge toward one another. By ending the lashing turns before the three apices touch, an. open space is left; and if the turns are continued after the apices touch, a central triangle is formed over the apices of the three triangles. All three-variations are used; they may be termed (1) open triple triangle, (2) closed triple triangle, and (3) overlapping triple triangle. (See figure 105.)
Figure 102.—Triple triangle lashing; commencement fixation. a, front: braid commencement end (1) laid over adz butt as it rests in bed on foot; braid runs obliquely upward to left, passes around back of heel, and appearing on right (2), crosses obliquely downward over braid end to fix it. b, right side: braid continues obliquely downward (3), passes under shaft where a notch prevents slipping and ascends vertically (4) on, opposite side of shaft to meet first loop (1) at heel angle. c, left: braid crosses over loop at back of heel, descends obliquely (5) above upper oblique turn (b,1) on right side of; foot, descends obliquely (5) on left side to cross first vertical turn (4) under shaft and ascends vertically (6) on opposite side of shaft. d, left: from position c,6; braid passes obliquely downward (7) below previous upper oblique (2), crosses front and opposite right side (7′) to cross previous vertical turn, and ascends vertically (8) on near side of shaft on right of previous vertical turn. e, front: showing last turn (7); diagonal turns effectively fix commencement and establish courses for following stage.
Note. If diagonal turns are continued alternately above and below previous turns, adz head will become firmly lashed and simple lozenge pattern develops. This is used with medium hafts in other areas.
Figure 103.—Triple triangle lashing; double loop technique. From oblique crossings, turns change to alternate lower and upper loops crossing in front. a, right: from position 8 in figure 102 d, braid (9), keeping below upper oblique turns on right side of foot, passes to front where it turns obliquely upward on left below upper oblique turns, crosses back (9) of shaft, and descends vertically (10) on left of previous vertical turn, b, front: first lower loop (9, 9) in which turn 9, instead of continuing obliquely downward, turns upward below upper oblique turns on left side of foot. c, right: from last position a,10, braid crosses under shaft, ascends above lower oblique turns on left side, loops over previous loop (9) in front, descends (11) on right side above lower oblique turns, crosses under shaft, and ascends vertically (12) on opposite side. d, front: shows first lower loop (9) crossed by first upper loop (11). e, right: from last position c, 12 a series of lower loops by technique a, b, and upper loop by technique c, d are made alternately so that two upper loops and three lower loops are added to first two (9, 11); last turn shown descending (13). f, front: showing first lower loop (9), first upper loop (11), and additional loops.
Note. If this technique is continued with double loops, loop crossings will continue to develop the one anterior triangle shown in e; but as braid in vertical turns around shaft will always be to the left, other triangles cannot be developed. This technique is used in Tubuai, Austral Islands (5, pp. 155, 156).
Figure 104.—Triple triangle finish. The object is to alternate the vertical turns around the shaft from left to right of the first vertical turn (6) in figure 103, e. This is done by dividing the upper loops into an ascending and descending stage. a, front: from last position (fig. 103, e,13) braid ascends to form an upper loop (14) but after crossing middle line in front, instead of completing the loop by descending, it turns off and ascends above upper oblique turns to back of heel. b, left: shows turn (14) ascending on other side and emerging at back of heel above upper oblique turns and in position to descend vertically. c, left: from last position, braid descends vertically (15) on left of previous vertical turns and thus commences upper and lower triangles by crossing upper and lower oblique turns; after crossing below shaft, braid ascends vertically (16) on opposite also on left of previous vertical turns. d, front: braid commences descending stage of divided upper loop by descending obliquely (17) above upper oblique turns, crosses ascending stage (14) in middle line in front. e, right: after crossing in front, braid (17) descends above lower obliques, crosses inner vertical turn (16), passes under shaft and ascends on opposite side (18) where it is again on outer side of vertical turns. f, right: from last position (e,18) braid makes a complete lower loop (19), reaches back of shaft (20), and descends vertically (21) to lower edge of shaft, where it is in position to make upper loop in two stages. g, right: braid crosses under shaft, ascends (22) above lower obliques on far side, and crossing in middle line in front ascends (22) above upper obliques on near side to back of heel, where it completes ascending stage of upper loop; it descends vertically on far side to right or inner side of previous verticals, crosses beneath and ascends vertically (23) to back of heel. h, right: from back of heel braid descends above upper obliques on far side to cross in middle line in front (24) over ascending stage (22), descends obliquely above lower obliques on near side, and after thus completing descending stage of divided upper loop, it ascends vertically (25) on far side to be again on left or outer side of vertical turns and in position to form a complete lower loop.
Note. The technique is continued by alternate complete lower loops and divided upper loops, thus forming upper and lower triangles in addition to anterior triangle in pattern of a "triple triangle." Method of disposal of braid end will be mentioned with individual adzes to be described.
Figure 105.—Variation of triple triangle pattern: a, open triple triangle; b, closed triple triangle where three apices meet; c, overlapping triple triangle, simply a continuation of b when middle triangle is developed and by overlapping conceals apices of the three triangles.
Figure 106.—Triple triangle pattern on working adzes: a , closed triple triangle (British Mus., + 1612), attributed to Rarotonga; c, overlapping triple triangle (Bishop Mus., 8040), Aitutaki. a, haft length 21.75 inches; foot length 4.25 inches, slight dorsal knob (1) on proximal end of shaft; heel with bilateral rounded projections (2), the lashing braid evidently had some fixation turns around the foot and adz butt and then made a couple of vertical turns over heel knobs, after which triple triangle pattern was developed; last vertical turn around shaft, when it reached top was passed around a crossing turn and carried on down to take part in lashing bast collar (3) to shaft; collar (3) formed of strips of hibiscus bast compressed into a roll on back and sides of shaft with free ends hanging down on either side for 94 mm., roll part being 16 mm. in diameter; the braid from adz-head lashing took some wide spirals around collar and some close transverse turns around both limbs below shaft. b, front of heel, shows lateral projection (2, 2) and crossing turns passing over them. c, haft length 15.5 inches, front height 4.25 inches. Haft has rectangular enlargement (1) on haft carved into triangle and lozenge patterns, also on shaft; pronounced heel (2) with median posterior edge and point end cut off; adz head with marked shoulder (3) typical of Aitutaki; pattern with overlapping triangle (4); last vertical turn on far side has braid end tucked in under upper horizontal turns (5); treatment of shaft is post-European.
In ceremonial adzes the shaft expands into a pedestal, while the foot proper and the heel are horizontal. In both working and ceremonial adzes the hafts are without a toe and thus conform to the heel-haft type. As the lashing technique was worked out on a ceremonial adz, the figures in the following diagrams are shown with the foot and heel horizontal instead of vertical. The shape of the foot as adjusted to suit the lashing is described here instead of under the haft. (See figure 107.)
Figure 107.—Foot of Mangaian ceremonial haft. a, left side: 1, foot proper; 2, heel; 3, shaft expanded to form pedestal; 4, constricted neck between shaft and foot; 5, upper surface forming straight or curved line; 6, front surface continued downward and forward to meet front edge (7) which slopes downward and backward and may be slightly concave; 8, back edge sloping downward and forward to neck; 9, back surface formed by cutting off heel point. b, upper surface (5): narrow at back, expanded in front where slot (10) is cut out to form bed for butt of adz head. c, front: shows triangular front surface (6) with slot (10) in base above for adz butt and apex below where it meets front edge (7) formed by inward slope of lateral surfaces. d, back: shows back triangular surface (9) due to cutting off heel, base above, apex below joining back edge (8) formed by inward slope of lateral surfaces. [The anterior surface (6) and posterior surface (9) may be curved.]
The adz butt is fitted to the slot on the upper surface, and usually a piece of shark skin is laid over the butt and held in position by the lashing, while a short cuff of the skin projects beyond the lashing. The shark skin gives a grip to the lashing turns, and the short cuff is a purely conventional feature taken from the upturned cuff in working adzes which protects the lashings from contact with the wood being worked.
The decorative introduction formed by oblique crossing turns is detailed in figure 108. It will be observed that some of the crossing turns pass over the slot carrying the adz head, but this is merely to fill in the area required and does not assist in fixing the adz head in position.
Figure 108.—Mangaiari triple triangle, preliminary decorative stage. In all figures front of foot is to left. Sennit turns are in two oblique series but will be referred to as vertical and horizontal. They follow similar courses on near and far sides of foot; on near side, vertical turns run upward and backward and horizontal turns backward, while corresponding turns on far side run in opposite direction. a, commencement end (0) is bent around front edge near neck; first turn (1) is vertical toward upper back corner; braid crosses small posterior surface obliquely downward (a′, 1) and makes a forward horizontal turn (dotted line) to front edge. b, braid forms first horizontal turn (1′) on same line as on far side, crosses posterior surface (b′, 1′) from below upward to cross upper surface reversing course of a, 1 and makes first vertical turn downward on far side to front edge, where it crosses commencement end (0). c, coming around front edge, second vertical turn (2) diverges to run parallel on left of first turn (1), crosses upper surface diagonally to pass obliquely across posterior surface (c′, 2), whence it makes second horizontal turn on far side above first horizontal on that side, and reaches front edge. d, crossing front edge above first horizontal, the turn (2′) crosses the posterior surface (d′, 2′), corner of upper surface, and makes second vertical on far side to reach front edge with required spacing above second near vertical (2). e, third near vertical turn (3) passes over upper surface diagonally to cross posterior surface (e′, 3) to right side of second vertical (2) and above it on far side; it then makes third far horizontal (dotted line) to reach front edge above second near horizontal (2′). f, third near horizontal turn (3′) passes backward to hook around near corner of posterior surface (f′, 3′) to form third far vertical which descends to front edge. g, fourth near vertical (4) passes up on left of third vertical (3) and loops over upper surface (g′, 4) and around far margin and lower apex of posterior surface (g′′, 4) to reappear on near side below first horizontal (1) to form fourth near horizontal (g, 4′) which crosses front edge parallel with first horizontal (1). h, braid forms fourth horizontal on far side below and parallel with previous horizontal, makes reverse loop (5), and descends on far side to complete fourth far vertical (5). [It will be noted that first three verticals near and far cross small posterior surface, and corresponding horizontals form on opposite side. The three posterior crossing turns from either side cover posterior surface, as shown in a-f. After this, turns take a complete loop around projecting heel, so that both verticals and corresponding horizontals are on same side, as shown in g, h. Horizontals, however, are now below preceding horizontals, so that verticals and horizontals converge toward each other to fill in uncovered space. First loops on either side of posterior surface set line of subsequent loops which must be parallel to first, and as loops get larger a third set of oblique turns are formed which cross between upper and back edges of heel and foot.] i, shows heel and foot covered; turns parallel with first loop (5) from far side are succending loops formed from far side while far side has corresponding loops formed from near side, as in g, 4.
Figure 109.—Mangaian triple triangle, final lashing stage. In figures, front of the foot is to right. Completed decorative stage is not drawn to avoid confusion. Courses followed by braid on lateral surfaces divide into four sets which, regardless of direction, may be conveniently termed front, back, middle, and lower. a, commencement end (0) is bent around back edge a little distance above neck; braid makes first middle turn (1) upward and forward to angle formed by front surface and front edge, turns upward on far side along edge of front surface to make first far front turn (2); on upper surface it makes an oblique turn (3) backward to near edge of upper surface (3, and a′, 3) whence it descends to starting point as first near back turn (4); it crosses and fixes commencement end (0), and on far side makes first lower turn (5) to front edge just above neck. b, this is the opposite of turns in a; from front edge braid returns on near side as first lower turn (6), crosses the back starting point, ascends on far side as a back turn (7), crosses upper surface obliquely forward (8 and b′, 8), descends on near side along lateral edge of front surface as a front turn (9), crosses front edge below anterior surface, and makes a middle turn (10) on far side back to starting point; four sets of turns are now completed on each side. c, an alternating series of turns now commences; from starting point braid crosses near lower turn and turns upward to make a back turn (11) on right side of previous turn (4); it crosses upper surface obliquely to middle line but, instead of continuing a straight course, curves backward to lie on right of previous turn (c′, 12 on right of 8). d, braid descends on far side as a back turn keeping on right (front) of previous turn on that side, reaches posterior edge below starting point, and forms a lower turn (13) below the first lower turn (6) (d′ shows position on upper surface). e, braid turns around front edge, makes a far lower turn keeping below lower turn on that side, crosses back edge, and ascends forward on near side as a middle turn (14) keeping below previous middle turn (1); from this ascending middle turn first series (a to b) is repeated. f, from middle turn (14), braid crosses front edge, turns up as a far front turn to left of previous one, crosses upper surface obliquely backward (f′, 15) to near side, whence it descends as a back turn (16) keeping on left of nearest back turn (4), and reaches back edge above starting point. g, from back edge, braid makes a lower turn on far side keeping above previous lower turns, crosses front edge, and returns as a near lower turn (17) above previous lower turns, and so crosses previous middle and back turns to reach back edge. h, from back edge braid turns upward on far side to form a back turn on left of previous turns, crosses upper surface obliquely forward (h′,18), and descends on near side as a front turn (19) to front edge. i, from front edge, braid runs on far side as a middle turn below previous middle turns and reaches back edge below previous crossings (20). This stage repeats the first series (a, b). The second series (c) is commenced by braid turning upward to form a back turn (20) on right of previous turns, and reaching upper surface it repeats previous loop (i′, 21). j, from upper loop braid descends on far side as a back turn keeping on right, and crossing back edge below previous crossings, it makes a near lower turn (22) below others. k, from front edge a lower turn is made on far side to back edge from which braid ascends forward as a middle turn (23) to front edge, keeping below the others. This middle turn commences series a, b over again. l, from front edge braid ascends as a far front turn, crosses upper surface obliquely backward (l', 24), crosses near upper edge, and descends as a back turn (25) on left of others; at back edge braid is above previous crossings, m, from back edge a lower turn is made on far side above previous ones; braid turns around front edge and returns on near side as a lower turn (26) which crosses previous middle and back turns on its way to back edge, thus forming commencement of two small triangles. n, from back edge, a back turn ascends on left of previous turns, crosses upper surface obliquely forward (n′, 27), descends on near side as a front turn (28), crosses front edge to cross on far side as a middle turn below others and so reaches back edge below previous crossings (29); this completes third set of first series (a, b). o, second series, with loop crossing above, is again repeated by turning upward from back edge as a back turn (30) on right of previous turns, and reaching near upper edge it loops backward over upper surface (o′, 31). p, braid turns down on far side to form a back turn on right of previous ones and reaches back edge (32) below previous crossings, when it makes a near lower turn (33) below the others, turns around front edge; returning on far side as a lower turn, it reaches back edge (34) where, haying completed a second series, it is in position to commence a first series with a middle turn on right of previous ones. q, by continuing first and second series of turns alternately, lashing is completed as shown in q; last turn is a lower turn (35) which, returning on far side to back edge, has braid end (36) tucked in under back turns.
Figure 110.—Multiple forms of Mangaian triple triangle: a, ceremonial adz (Cambridge University Mus., 1903-282), collected in 1848; first series (1) commences some distance away from front edge, a second series (2) is established by a shift forward, and another forward shift establishes finishing series (3) with its upper triangle near front edge. b, upper surface of a showing crossing of the three series (1, 2, 3); c, back of a, also showing overlapping of three series (1, 2, 3); d, ceremonial adz (?) showing first series (1) and backward shift of finishing series (2). To avoid confusion, some lashing turns are not filled in in c and d.
I examined the following 16 utilitarian adzes hafted to plain hafts: British Museum (9305, 1925-196, 1904-256), A. W. F. Fuller collection (14-11-19, 14-11-29, 272, 271, 19-11-29), Oxford University Museum (1915), Cambridge University Museum (Z, 6093), Liverpool Museum (4996-M), Peabody Museum, Salem (E. 5119, E. 15807, E. 22, 452), and Wesleyan University Museum, Middletown, Connecticut (344, 373). The total length including shaft and foot ranges from 20.5 to 26.5 inches. Some have a median edge on the upper side of the shaft and two have the proximal ends of the shaft carved for a short distance. All have characteristic Mangaian adz heads of polished black basalt with a high bevel apex which reaches the butt. In four there are bilateral rounded projections at the top of the haft heel resembling the form, in the adz attributed to Rarotonga (fig. 106, a,2). The lashing pattern in nine is the normal Mangaian triple triangle; in six the multiple form is produced in series of three in four and series of two in two; in one the overlapping form is present. In all, the preliminary decorative oblique turns are applied before the true lashing. In two, the original lashing in narrow braid 2 mm. page 169wide is covered with a later lashing in wider braid 3.5 and 5 mm. wide. Examples of single, multiple, and overlapping patterns on working adzes are shown in figure 111.
Figure 111.—Mangaian lashing patterns on working adzes, a, single pattern (Cambridge University Mus., Z.6093): single front triangle (1) with other two (2, 3) in braid 3.5 mm. wide worked over original similar pattern in narrower braid; total length of shaft and foot 678 mm., shaft proximal end diameter 36 mm.; slight neck constriction (4) at shaft and foot junction, b, multiple pattern (British Mus., 1904-256): front triangle of first series (1), second series (2), and finishing series (3); heel has rounded lateral projections (4). c, overlapping pattern (British Mus., 9305): lashing, after developing usual pattern in a, makes additional turns which result in enlargement of three triangles (1, 2, 3) and apex of third triangle (3) not only reaches front edge but is overlapped by succeeding turns which cover apex of 3 and form extra triangle 4. Note the shaft and foot constriction (5) and rounded lateral projections (6) of heel which have been broken off. Last descending turn (7) crosses middle line and is tucked under a crossing turn; decorative crossing turns (8). Total length of shaft and foot 23.25 inches, d, back of heel of c showing broken lateral knobs (6, 6) and tucking down of last turn (7).
The use of the Mangaian form of the triple triangle pattern on so many authentic working adzes proves that this form was the current technique and was not confined to ceremonial adzes. Working and ceremonial adzes have a similarly shaped foot, which was evidently developed as the form best suited to display the purely decorative part of the lashing. Lateral lugs on the heel in four out of 16 adzes also establishes this form of heel for Mangaia and page 170thus makes the adz haft attributed to Rarotonga (fig. 106, a) liable to query. The haft appears to be Mangaian and old, whereas the adz head and lashing are probably Rarotongan. It may be that Gill had an old haft from Mangaia which was fitted with a local adz head in Rarotonga and lashed with the Rarotongan lashing technique.
A number of hafted adzes variously attributed to the Hervey Islands and Tahiti show a distinct departure from the technique of the two patterns just described. It is evident that a series of spaced lashing turns were so arranged that the crossings resulted in an outline arrangement of a number of lozenge-shaped figures. These lozenges were then filled in by successive lashing turns and the technique of filling consists of two different methods. As the two techniques have already been described for the house rafters of Mangaia (fig. 18) and the fan handles of Rarotonga (fig. 23), I will term them as (a) the Mangaian multiple lozenge and (b) the Rarotongan multiple lozenge. Even if later studies should prove that the adzes lashed with the multiple lozenge techniques belong to other islands, such findings need not alter the technical terms employed here.
In the Mangaian multiple lozenge pattern, the technique is identical with the inaere pattern used in Mangaia on house rafters (fig. 18) and the stand of a wooden god (fig. 240, d). This technique has not been recorded for any other island in the Cook, Society, and Austral Islands. Four adzes lashed with this technique are distributed as follows: two in the British Museum (55/12-50/172 and L.M.S. 382) and one each in the Liverpool Museum (18-12-77-8) and the Auckland Museum, New Zealand (1310). All the adz heads are of the inverted triangular type with dropped shoulders. All are working adzes with old sennit lashings and shiny shafts denoting age. The L.M.S. specimen in the British Museum is smaller than the others and has a straight shaft but the three larger ones have a bend in the shaft (pl. 9, C). Another adz in the A. W. F. Fuller collection (lot 321) has a similar bend in the shaft but the foot and adjacent part of the shaft are completely covered with a lashing of twisted pandanus leaf. I feel that the pandanus leaf lashing is merely a covering to protect an original sennit lashing in the Mangaian multiple lozenge pattern which lies beneath.
As in the inaere rafter pattern, a series of turns are made with the lashing braid to map out the crossings to form the lozenge figures. The crossings are possible over the foot and the adjacent shaft, but there is a triangular space over the junction of foot and shaft where crossings are not possible. The outline of the first series of turns is shown in figure 112, a. Subsequent series of turns follow the course of the foundation series, but each series is immediately above the preceding series. In other words, each series is on the same side of the foundation series. This technique is followed until the lozenges are filled in (fig. 112, c).page 171
Of the four adzes with the Mangaian multiple lashing, the small one from the London Missionary Society's collection is attributed to the Hervey Islands, two are attributed to Tahiti, and one is unlocated. Tahiti may be disregarded, as the inaere lashing has not been recorded on artifacts from that area and the adz heads are better finished than those of Tahiti. Three of the adz heads are made from the close grained black basalt so characteristic of Mangaia and their shape agrees with Mangaian forms. The presence of the inaere pattern in Mangaia adds further evidence that this form of adz lashing may belong to Mangaia. The inaere pattern occurs in Fijian adzes lashed to a long toe, but the adz head form and the haft without a toe belong definitely to the central Polynesian area. The high heels of the hafts are a departure from the pattern observed in the other Mangaian adzes, but it is evident that this form of heel was better suited for working out the multiple lozenge pattern. Though the Mangaian triple triangle was the recognized form of lashing for ceremonial and working adzes, it would appear that some craftsman followed an artistic urge in using the inaere technique as an adz lashing.
Figure 112.—Mangaian multiple lozenge lashing pattern. a, length of shaft and foot, 26 inches; height of foot, 6.3 inches; high heel with no lateral or posterior projections; shaft proximal end diameter, 1.2 inches; long narrow adz head with posterior median edge rounded off, marked shoulder, conforms to a Mangaian type; some ornamental turns of braid pass over end of heel and then lashing in series of spiral turns forms lozenges, which are not filled in the figure (British Mus., 55/12/50-172, Tahitian case). b, top of heel of a forming triangular surface with apex (1) to back and notches on sides to hold decorative turns passing over it, as in Mangaian technique. c, small specimen, haft length, 13.5 inches; shaft diameter near proximal end, 13 mm. but expanded at end to 22 mm., well polished; adz head of black basalt: blade length, 65 mm.; width at shoulder, 32 mm., at cutting edge 36 mm.; thickness at shoulder 17 mm.; conforms to a Mangaian type; lashing in multiple lozenge pattern and a couple of turns pass over top of heel (British Mus., L.M.S. 382).
Figure 113.—Rarotongan multiple lozenge pattern (W. O. Oldman coll., 470). a, front showing lashing and upper part of adz head: 1, dropped shoulder of adz head; 2, median edge on butt; 3, top of adz head with two rounded nipples; 4, heel of haft, b, side view: showing multiple lozenges extending onto shaft.
Many Mangaian hafted adzes have completely carved shafts of the size ordinarily used in woodwork. I regard such adzes as post-European and made for trade. In the series of 16 adzes with plain shafts that appear to be old, two have the proximal end expanded and carved with the K-pattern page 173characteristic of Mangaia. The carved ends of the two shafts are shown in figure 114. An adz from Aitutaki with a carved enlargement on the shaft toward the foot is shown in figure 106, c. Such enlargements occur on other Aitutaki adzes, but I also regard them as post-European.
Figure 114.—Carved adz shafts, Mangaia. a, length of carved part, 62 mm., gradually expanding from diameter of 30 mm. to 44 mm. at proximal end; carved with Mangaian K-motif in three transverse panels around shafts, followed by longitudinal panels with slits between them, one more transverse panel, and finally short knobs at proximal end (A. W. F. Fuller coll., 19-11-29). b, rectangular enlargement of proximal end, 1.87 inches long by 1.62 inches wide; carved in three transverse panels with K-motif (Bishop Mus., C9379).
Sennit ornamentation of shafts
Of the series of 16 Mangaian adzes with plain shafts, no less than nine are ornamented with sennit lashings around the shaft. However, they were acquired by collectors for their more artistic appeal and cannot be regarded as an indication of what prevailed in the normal technique of every day life. The forms of shaft lashing are: spaced transverse bands, continuous zigzag, and continuous ellipses (fig. 115).
Figure 116.—Technique of ornamental lashing. a, transverse spaced bands: commencement end (x) is upward toward foot of haft, held with left thumb, and four loose turns (1-4) are made around shaft from right to left; free end (5) of braid is passed under all four turns, and each turn is drawn taut successively from below up (1-4); slack which accumulates with last turn (4) is then drawn down under turns by pulling on free end (5). The technique buries commencement end (x) and leaves free end of braid to pass downward and form another band below with same technique. b, zigzag pattern: looking down on dorsal surface of shaft. Commencement end of braid is fixed by a couple of transverse turns below adz lashing and braid makes its first course (1) by spiral turns around whole length of shaft, care being taken to keep turns parallel. On dorsal surface, this first course (1) is shown descending from right to left but on under surface it descends from left to right as shown by dotted line. At end of shaft, braid takes a transverse turn and then ascends with second course (2) to cross first course (1) in middle of dorsal and under surfaces and so sets up pattern to be followed by subsequent four courses. Note that on dorsal surface second course (2) ascends from right to left, the reverse on under surface. Third course (3) descends and keeps on upper side of first (1) while fourth course (4) ascends on lower side of second course (2). Having reached top, fifth course (5) has to descend on lower side of first (1) because upper side is already occupied by third course (3). From bottom, sixth and last course (6) must ascend on upper side of second course (2) because lower side is occupied by fourth (4). End of braid may be fixed by passing it under a transverse turn and cutting off slack. A little adjustment is needed at the upper and lower ends of shaft, but once first two courses are established, others follow automatically in sequence. It really does not matter whether third course passes above or below first, or fourth above or below second. Fifth and sixth courses complete pattern by taking unoccupied side. c, d, e, elliptical pattern: viewed from above but with under surface spread out on flat; two middle ellipses are seen from above and two outer ellipses meet in middle line of under surface. c, braid commencement end (x) is turned upward on left of dorsal middle line and after four loose turns (1-4) are made from left to right around shaft, free end (5) is passed downward below them on right of mid line. In the figure each turn has a right and a left half. The first turn (1, 1) is drawn taut to mid line and then second (2, 2) turn, which fixes the commencement end (x). Then right half of third turn (3) describes a curve below first turns (1, 2) in inner right quadrant and above them in outer right quadrant, the slack of left half (3′) being shown still slack, d, from last upper curve on right, left half of third turn (3′) describes a lower curve in outer left quadrant and commences an upper curve in inner left quadrant. It cannot curve down to make a compensating lower curve in adjacent inner right quadrant because that curve is already occupied. Third turn (3′), therefore, passes straight across mid line where it becomes right half of fourth turn (4) which turns down to make a lower curve in outer fourth quadrant. Slack of fourth turn (4′) is shown on left where it is above first two turns. e, slack of fourth turn (4′) now completes an upper curve in outer left quadrant, and completes a lower curve in inner left quadrant. Accumulated slack of fourth turn is removed by pulling on free end (5) and this by pulling down fourth turn (4, 4′) in middle line completes upper curves. Thin flat braid enables curves to retain their position.
The spaced transverse bands form the commonest motif, six out of the nine adzes being so treated. The bands consist of three or four close turns of sennit which are made in a spaced series of two to four bands around the proximal (2) or distal (2) ends of the shaft. One adz has four bands at the distal end and two at the proximal, whereas another has a number of bands around the middle of the shaft. Each series is continuous in that the braid goes on to the next band without being cut. A similar technique was used on the shafts of some Samoan and Tongan clubs and in such weapons, the bands served a useful purpose in giving a firm grip to two hands. While such a use may be attributed to the bands around the proximal ends of the adz shafts, their presence on the distal ends indicates that what may have been useful originally became on some shafts a purely art motif. In none of the six adzes are the bands continued over the whole length of the shaft. The details of technique are shown in figure 116, a.
The continuous zigzag pattern occurs on one adz and covers the entire length of the shaft. The technique is similar to that used on the Mangaian fan (fig. 25), Mangaian breast ornament (fig. 63), and Mangaian rafter pattern (fig. 17, e). The order in which the successive courses are made is shown in figure 116, b.
Figure 117.—Triangular adz, intermediate; Rarotonga (C8880). a, front: long blade, narrow cutting edge; dropped shoulder, b, back: small bevel surface, straight transversely; median edge extends to poll. c, right side: butt slants backward. d, sections. Weight, 3.25 oz.
The continuous ellipses occur on two adzes: A. W. F. Fuller collection (271) and Peabody Museum, Salem (E. 22, 452). The lashing is made with fine sennit (rapa) two mm. wide and one coir fiber in thickness. The ellipses are arranged in bands of four which are continued with the one braid over the whole length of the shaft. (For technique, see figure 116, c-e.) The two adzes so treated are beautiful works of art.
Distribution and Evolution
For the distribution of haft types and the evolution of lashing patterns, see pp. 442, 449).
Gouges and Chisels
In the Bishop Museum collection there are no gouges, but it is possible that the adzes with a concave bevel surface may have served as gouges.
It is also possible that some of the implements with a narrow cutting edge (fig. 117), though hafted like adzes may have performed the work usually done with a chisel. In Samoa, I saw metal chisels hafted like adzes being used by the native carpenters. The only tool in the Bishop Museum collection that could be regarded as a chisel (fig. 118) is triangular in section with a narrow cutting edge. A chisel from Aitutaki is also roughly triangular in section (70, p. 241).
Figure 118.—Chisel, Mauke (C2701). a, front: butt and blade in one plane, no shoulder; lateral edges somewhat rounded; blade narrows to cutting edge. b, back: bevel surface straight and merges into back without distinct apical point, because median edge is rounded off; rounded median edge extends to poll. c, right side. d, sections. Weight, 3.5 oz.