A.—Adzes Triangular in Cross Section.

(1.) Triangular in cross section, with base anterior.

This division forms by far the commonest type in the Cook Islands. Linton says they are characteristic of the south-eastern Polynesian area. Of the six Aitutaki adzes, five were of this type, and in Skinner's five three were triangular.

In the characteristic form the base of the sectional triangle lies anteriorly. The three main surfaces of the adze are thus anterior and right and left postero-lateral.

Figure 193.
Aitutaki Adzes, triangular in section, with long blades.

The anterior surface is definitely divided by a well-formed butt shoulder into the lower quadrilateral blade surface and the upper butt surface, which is shaped into a grip. Posteriorly, the postero-lateral surfaces meet in a well-defined median posterior longitudinal edge in the upper part of the adze, and are separated in the lower part by a well-defined triangular bevel surface.

The above general characters can be seen in the adzes figured, which are shown with an anterior, posterior, and right lateral view. This order is observed throughout. In some of the series of three views the adze is not always of the same size, owing to its having been reproduced from photos. taken in different focus.

The butt is shaped to form a grip by working the anterior surface and the adjoining parts of the posterolateral surfaces. The longitudinal edges bounding the anterior surface of the butt are thus rounded off. The posterior longitudinal edge is well defined and continuous with that of the blade. The poll may be rounded or almost pointed, but it is usually a well-defined superior triangular surface that may be chipped level, or even ground.

The butt shoulder is formed by the shaping of the anterior surface of the butt and the rounding off of the longitudinal edges. It is a characteristic feature of these adzes.

The blade has a quadrilateral anterior surface, bounded laterally by the longitudinal edges, above by the butt shoulder, and below by the cutting edge. It is usually wider at the cutting edge, but it may be of even width throughout, or even narrower at the cutting edge. The anterior surface has both a longitudinal and a transverse convexity. The former is accentuated near the cutting edge, and the latter near the longitudinal edges. The postero-lateral surfaces of the blade are continuous with those of the butt, except in the anterior parts of the latter that have shared in the shaping of the grip. The surfaces are convex antero-posteriorly.

The triangular bevel surface is another characteristic feature. The base of the triangle is formed by the cutting edge, and the sides by the oblique edges where the bevel surface meets the two postero-lateral surfaces of the blade. The apex is formed by the three surfaces meeting at the posterior longitudinal edge. This point regionally represents the bevel shoulder of quadrangular adzes. For convenience, it has been referred to as the bevel apex. The bevel surface is usually concavo-convex. It is concave transversely at and near the cutting edge, where it follows the transverse convexity of the anterior surface of the blade. It may be a marked feature that extends posteriorly on the bevel surface. It is due to extra grinding in the mesial line with a hone. See the cross sections through the bevel surface in Figs. 196 and 198. When the bevel surface is ground level transversely there is a downward convexity at the cutting edge. The convexity of the bevel surface is longitudinal and is accentuated near the cutting edge. See page 223the right lateral views in the various figures, where the bevel surface is on the left.

Figure 194.
Aitutaki Adzes, triangular in section, with shorter blades.

Variations. The above description applies to most adzes of this type, but there is great variety, not only in size, but in the proportion of the various parts to each other.

(1.)	Longitudinal axis of butt and blade. Instead of the butt being in the same longitudinal axis as the blade, it may have a posterior inclination of varying degree. This accentuates the general anterior convexity of the adze, Fig. 194A.
(2.)	Relative length of butt and blade. The butt may be short as compared with the blade, Fig. 193. In long-bladed adzes, it is natural for the butt to appear relatively short. When the butt reaches its maximum length of utility, it is unnecessary to keep on increasing it in proportion to the increasing length of the blade. In spite of this, however, some butts are shorter with long blades than in others with blades of lesser length. The dimensions of the adze in Fig. 193B are shown in Fig. 196. The butt may even be longer than the blade, Fig. 194B and 195. It is natural that with short blades the butt should be relatively long. There must be sufficient length of butt to provide secure hafting. In many adzes the relative shortness of the blade is due to re-grinding the bevel surface, to sharpen or to get rid of gaps. An early writer states that the wood-workers in Tahiti had water and a sharpening stone beside them, and they were constantly re-sharpening their tools. Every page 224 Figure 195. Aitutaki Adze with butt longer than blade. time an adze is sharpened the blade is shortened. The time must sooner or later arrive when the adze blade is too short to be useful. It is cast aside as a reject. Years later, when the villages of the stone age period have disappeared, it is picked up and gets into circulation, to cause conjecture amongst archaeologists. Each one has to be examined carefully before it can be decided whether it came from the refuse heap or the tool shop. Probably an occasional reject escapes the censor and enjoys the unsought honour of figuring as a new type.
(3.)	Relative width of the butt shoulder and the cutting edge. (a.) Most adzes are wider at the cutting edge. The amount of increase in width per length, however, varies considerably, and naturally alters the appearance of the adze. Fig. 194A. Figure 196. Aitutaki Adze, long blade, narrower at cutting edge, bevel surface, concave transversely. page 225 (b.) The width may. be the same throughout, Fig. 194B and 195. (c.) The cutting edge may be narrower, Fig. 196. This is by no means an uncommon feature. Of 40 Rarotongan adzes examined for this feature, 13, or 32 per cent., were narrower at the cutting edge. This feature usually occurs in the long, narrow adzes, with a marked transverse concavity on the bevel surface. The above characteristics are shared by the long Aitutaki adzes, Fig. 19, and the unground Rarotongan adze, Fig. 197A. For comparison, the principal measurements are given as follows:— Length Width Thickness Length Total Butt Blade B.S. C.E. B.S. Bv.S. Aitutaki, Fig. 193B 150 45 105 39 33 33 75 Aitutaki, Fig. 193A 250 80 170 51 46 43 93 Rarotonga, Fig. 197A 264 100 164 56 55 57 104 B.S.—Butt shoulder. C.E.—Cutting edge. Bv.S.—Bevel surface. The thickness is taken just below the butt shoulder, and the length of the bevel surface from the cutting edge to the bevel apex. Figure 197. Rarotongan Adzen, not quite finished by sharpening or polishing. The other adze, Fig. 197B, is also unground and unpolished. It has a marked butt shoulder ridge, and the blade is wider at the cutting edge. page 226
(4.)	Proportion of width of blade to length of blade. The proportion is naturally noticed best by a view of the anterior surface of the blade. The blade may be long and narrow, Fig. 193, or short and wide. Fig. 194. If the half of the added width of the blade at the butt shoulder and the cutting edge is taken as the mean width, we may get a blade width index expressing the proportion of width to length. Thus in the long narrow blade in Fig. 196 it is 34.2, whilst in the short stumpy blade in Fig. 195 it is 78.
(5.)	Treatment of the butt. A characteristic feature of the triangular adzes is the shaping of the butt to form a grip. This is done by working down the anterior surface of the butt. Whatever the primary processes may be, the surface is left roughened by grinding to give better support to the lashing, or any material that may be laid over the surface before the lashing is applied. The working of the anterior surface of the butt places it on a different plane to the anterior surface of the blade, and thus creates a butt shoulder. This assists in preventing the blade from working up under the lashing when a blow is struck. That the butt shoulder came to be regarded as more important than the roughened surface is indicated by the fact that in some triangular adzes, Fig. 194A, the anterior surface of the shaped grip is polished equally with the blade. It may be that in the method of hafting to be described, the roughening of the anterior surface of the grip was not necessary to give additional security to the lashing. In good adzes, the grinding down was primarily to provide a butt shoulder, and the surface was left rough, not to give security to the lashing, but because it was concealed by the lashing. Some more energetic craftsmen polished the surface. As the butt continues the triangular cross section of the blade and. the base of the triangle forms the anterior surface, the grinding down of the anterior surface naturally reduced its width. The reduction in width was further increased by rounding off the longitudinal edges that are formed on either aide with the postero-lateral surfaces of the butt. The shaping and grinding on these latter surfaces extended posteriorly for varying distances. It is usually just enough to round off the longitudinal edges. The posterior parts of the postero-lateral surfaces, as far as the posterior longitudinal edge, share in the polish of the page 227same surfaces of the blade. They are continuous with them. There is no ridge or shoulder to break the continuity. In polishing the upper part of these surfaces on the blade, the rubbing of necessity encroached on the butt surfaces, and the craftsman made a job of it by continuing the polishing to the poll. It is rarely that the grinding or roughening extends over the entire postero-lateral surfaces of the butt, but such occurs in the large Aitutaki adze, Fig. 193A. Variation occurs in the amount of grinding immediately above the butt shoulder. In some cases it is slight, and the surface slopes back gradually to the plane of the anterior butt surface. In other examples a groove is ground, which may extend slightly on to the postero-lateral surfaces. Again, this groove may have a downward inclination as in Figs. 195 and 196. This serves to throw out the corners formed by the butt shoulder with the longitudinal edges of the anterior surface of the blade. It is this appearance in adzes from other regions that gave rise to the term shouldered celt. The butt normally narrows towards the poll. In some cases the narrowing is more accentuated than in others. Thus in triangular adzes the butt is narrower than the blade, and suggests the use of the term tang. Attention is again drawn to the fact that this appearance is primarily due to grinding away of the anterior surface of the butt, which forms the base of a triangle. If, in the so-called tanged adzes reported from other areas, such as Cambodia, the appearance is primarily arrived at by working away the lateral surfaces, the method of manufacture must be considered in drawing comparisons. So also must the cross section of the adze.
(6.)	Treatment of the butt shoulder. (a.) Absence of butt shoulder: Adzes without a distinct butt shoulder form the exception. They are usually rough, and have an unfinished appearance. Such a one is the Rarotongan adze on the extreme right of Fig 199. (b.) Butt shoulder from shaping the grip: Attention has been drawn to the variations that occur in the grinding of the surface immediately above the butt shoulder.page 228 (c.) Transverse ridge on blade: In a fair number of adzes the upper margin of the anterior surface of the blade has been ground so as to leave a transverse ridge, Fig. 197. In the examples seen, this was always associated with the shaping of the grip. It further accentuates the butt shoulder, and thus further aids in preventing the adze from slipping up under the lashing of the handle. The transverse ridge without any shaping of the butt is seen in the quadrangular adze, Fig. 203.
(7.)	The bevel surface. The bevel surface is usually approximately straight transversely, but is generally concave at the cutting ridge. This tranverse concavity may be very well marked and extend over the bevel surface as far as possible. See cross section in Fig. 196.
(8.)	Posterior longitudinal ridge.

In the usual type of adze, the blade is longer than the bevel surface. The bevel surface thus meets the posterior longitudinal line of the blade at some point below the level of the butt shoulder. See posterior view of Fig. 196. A cross section of the blade above this point thus shows a typical triangle, with the apex formed by the posterior longitudinal edge.

In some cases, where the blade is short and thick, the bevel surface may be longer than the blade, Fig. 195. In other words, the bevel surface does not meet the posterior longitudinal edge on the blade at all, but at some point above the level of the butt shoulder, and thus on the butt. A cross section of the blade, taken even at the shoulder, passes through the bevel surface. The section, instead of being triangular, is thus quadrangular, Fig. 195. The section of the butt, however, shows the apex formed by the posterior longitudinal edge, but the other two angles of the triangle are rounded by the shaping of the grip. The adze, however, is so obviously of the triangular type that it may be treated as a variation, and not a sub-type. It is important, as it indicates what may happen to form a subtype.

In Fig. 198, the bevel surface would practically reach the posterior longitudinal edge at about the level of the butt page 229shoulder, but there is no longitudinal edge. The posterior longitudinal edge has had the edge ground away, and is thus represented by a narrow posterior surface. This has been partly due to the posterior inclination of the long axis of the butt from that of the blade. The result has been that in grinding the upper part of the bevel surface to the required plane, the posterior longitudinal edge has been ground for a longer distance than usual. The adzemaker has then finished the adze off by rubbing out the rest of the posterior longitudinal ridge. The type adze is shown

Figure 198.
Aitutaki Adze with posterior longitudinal edge ground away.

in Fig. 194A, whilst the dimensions and cross sections are shown in Fig. 198. Though slightly quadrangular, both in the blade and the butt, the adze is still so obviously of the triangular type that it remains a variation and not a distinct type.

The triangular adze has been described by Linton as the south-eastern type, "characterised by a triangular or subtriangular cross section, a relatively thin blade with a long bevel and a more or less distinct shoulder where the blade meets the tang."

Skinner, in his classification of adze types, has placed the triangular adze in Type V. To admit Cook Islands adzes into this type the words "undeveloped grip" as a characteristic feature would have to be altered.

Fig. 199 shows a number of well-made Rarotongan adzes that are triangular in section. The variation already alluded to is very obvious. The adze on the extreme right has no shoulder or grip, and is the only poorly-made adze of the series.

A—Anterior. B—Posterior. C—Right Lateral.

Figure 199.
Rarotongan Adzes, triangular in section. (Auckland Museum.)

A good example of a triangular adze from Mangaia is shown in Fig. 200A. It has a broad blade, and the grip shows more shaping than usual. Roughening also extends over the whole of the postero-lateral surfaces, except for a narrow strip at the posterior longitudinal edge.

The adze Fig, 200B, has a long blade, which narrows to 6mm. at the cutting edge. The anterior surface is well defined, as well as the butt shoulder and grip. Posteriorly, page 231

Figure 200.
Mangaian Adzes. (Auckland Museum.)

A—Typical triangular type. B—Long, narrow, deep blade.

the mesial posterior longitudinal edge has been ground off to form a narrow posterior surface. The marked feature is the narrow cutting edge and the depth at the butt shoulder. A similar one in the author's possession is mounted in a ceremonial peace adze handle. The stone is very black and highly polished. The type may be ceremonial. The measurements in millimeters are:—

			Length	Width	Thickness
	Butt	Blade	Total	B.S.	C.E.	B.S.
Mangaian Adze, A -	52	124	176	61	62	40
Mangaian Adze, B -	47	157	204	27	6	38

B.S.—Butt shoulder. C.E.—Cutting edge.

(2.) Triangular in cross section, with base posterior.

An unfinished Rarotongan adze in the Auckland Museum is shown in Fig. 201. It is the largest in the collection. There is no butt shoulder or grip, though something might have been done when the implement was ground down. The adze is triangular in section through the butt and the blade, the base being posterior. The axis of the blade is convex longitudinally. Towards the lower end, the anterior longitudinal edge, which forms the apex of the triangular section, broadens out into an anterior surface which runs out to the sides of the cutting edge. This surface is thus triangular, and extends for 124mm. page 232

Figure 201.
Uncommon Rarotongan Adze, with reverse triangular section.

from the cutting edge to its apex. The posterior surface is flat transversely, and is defined by two lateral longitudinal edges, the poll above and a bevel shoulder below. The surface is markedly concave longitudinally. This posterior concavity is more marked than the anterior convexity, and thus ends in a prominent bevel shoulder. See lateral view in Fig. 201. The width at the bevel shoulder is 63mm., and the quadrilateral bevel surface narrows to the cutting edge, which is 45mm. in width. From cutting edge to bevel shoulder it is 122mm.

The total length of the adze is 337mm. The width at the poll is 49 and the thickness 54. At the bevel shoulder the thickness is 63mm.

When trimmed up, the adze would have probably corresponded to the Marquesan toki kouma type described by Linton². It is probable, also, that in sharpening the cutting edge might have been rounded off, and thus corresponded to the Maori triangular adzes described by Best,³ which page 233he considered performed the functions of a heavy gouge, though hafted as an adze.

A broken specimen from the Kermadecs, now in the Auckland Museum, also shows the base of the triangle posteriorly. The grip is well formed and the anterior mesial ridge rounded off. The blade is well polished. There is an anterior projection at the poll.