Volume 49 1940 > Volume 49, No. 195 > The function of latte in the Marianas, by Laura Thompson, p 447-465
                                                                                             Previous | Next   

- 447

HIDDEN in the jungle along the shores and water-courses of the Marianas islands and scattered over the best garden-land are the ruins of hundreds of capped stone upright sites, called latte by the modern Chamarros. The most conspicuous archaeological survivals of the ancient native culture, these latte have been found on all the Marianas islands which have been examined. H. G. Horn-bostel (9), former Bishop Museum collector in the Marianas, estimates that there are about 270 latte sites on the island of Guam alone. Their approximate distribution and density are shown in figure 1.

Throughout the group the latte have been built according to a uniform pattern, a double row of stone up-rights, each surmounted by a hemispherical stone cap (fig. 2). In the course of time the sites have been disturbed by earthquakes and by the natives. Most of the caps have fallen from the uprights, some of which have been disturbed, and many stones have been removed (pl. 1).

There has long been a question as to the function of the latte which, on account of their apparent resemblance to other types of stone-work in the Pacific, have been considered by various writers as burial-monuments, temple-sites, or house-foundations. In this study an attempt has been made to determine their function by examining archaeological, historical, and ethnographical data on the Marianas islands obtained in Guam and in the Bishop Museum, Honolulu. Other sources may yield further information on the problem.

- 448

The following brief description of latte in the Marianas islands is based on personal examination of several latte on Guam, supplemented by data from the field-notes of H. G. Hornbostel (9), who has examined and described twenty-two sites on Guam, Rota, Tinian, and Saipan. For a detailed description and illustrations of type latte (see 14, pp. 8-20).

Classified according to length of ground-plan, there are three types of latte, namely, small, medium, and large. A small latte consists of eight capped stone pillars, of uniform size and shape, set in two straight parallel rows of four units each, according to a regular rectangular ground plan (fig. 2). Measuring from the centre of the base of each upright, the site is usually about 11½ feet wide and the same distance between the uprights in each row; that is, 34½ feet long. So far as can be determined from the ruins, the capped uprights of each latte on level ground were originally of uniform height, whereas those on slopes or uneven terrain were built up to a horizontal plane. Hence the height of the capped uprights in any one site tends to be constant. In latte of various types, however, it ranges from three to sixteen feet.

A medium sized latte consists of ten capped stone pillars set in rows of five units each, according to a rectangular ground plan of approximately the same width and distance between the uprights as in the small latte. The length of the site has been increased to about 46 feet, by adding a pair of capped uprights. A large latte consists of twelve capped stone pillars set in rows of six units each (fig. 3). Its width is the same as in smaller sites, but its length has been increased by still another pair of capped uprights to about 57½ feet. Thus latte increase in size according to a definite pattern in two directions only, namely in height of capped uprights and in length of ground plan, while the width and distance between the units is constant.

The material and form of uprights and caps, while uniform in a single site, vary between the islands and within each island according to the available material and the local style (14, fig. 3, 6, 7, 8). Latte located along the coast and elsewhere on limestone formation usually consist of uprights composed of limestone. They are frequently long

- 449
FIG. 1.
Island of Guam: showing distribution of latte.
- 450
FIG. 2.
Latte site on Marianas.
- 451
FIG. 3A., FIG. 3B.
Plan of large latte.
- 452
FIG. 4.
Latte parallel to shore.
- 453
FIG. 5.
Latte parallel to water-course.
- 454
FIG. 6.
Latte with axis perpendicular to shore.
- 455
FIG. 7.
Uprights (haligi or halege) of latte, set in ground.
- 456
FIG. 8.
Uprights of latte (haligi) with cap (tasa).
- 457
FIG. 9.
Miscellaneous burials between rows of uprights (haligi) of latte.
- 458

narrow slabs or pillars capped with inverted coral heads. Some are roughly finished, while others are carefully shaped. In the volcanic regions both uprights and caps have been laboriously hewn out of local rock to a form somewhat resembling those made of limestone, which are probably the proto-type.

Most of the latte are built with long axis parallel to the shore line (fig. 4) or water course (fig. 5), but some along the shore are oriented with axis perpendicular to it (fig. 6). As a rule single latte are isolated, but a number of sites, consisting of two or more latte, have been reported, especially from Tinian and Saipan (14, pp. 16-19).

The uprights (called haligi or halege) are set deeply in the ground and reinforced below the surface by stone props (fig. 7). Each cap (called tasa) is fitted to its upright by means of a slight cavity or groove on the lower surface of the cap (fig. 8). Frequently extended human skeletons together with miscellaneous human bones, potsherds, and broken stone implements have been found buried between and in the vicinity of the two rows of uprights (fig. 9). Skeletal remains and artifacts are also found below the surface between the latte and the shore. Moreover rough stone mortars, used by the ancient Chamorros as well as the modern Guamese for hulling rice (15), and large numbers of artifacts are found on the surface in or near practically every site.

Examination of the archaeological data alone suggests that the latte functioned as house-sites. Actually many of the facts can be explained satisfactorily only by this hypothesis. Strong uprights have been set deeply in the ground and reinforced with stone props, and the caps have been carefully fitted to them. Thus the capped pillars are unusually firm, as they would be to serve as house-posts. Moreover, as far as can be determined, the pillars were built up to a horizontal plane, as they would be to support house-beams. Furthermore, increase in size is achieved in length and height only, while the width is constant, a characteristic of Oceanian house-types in which the width of the building is limited by the length of the wooden crossbeams. Finally, artifacts have been found on the surface of many sites, which would be expected if they were used as - 459 house-foundations, since frequently the area under raised pile-houses is used as a workshop and also a refuse heap; and formal extended burials with broken implements have been found below it, indicating house-burial if the latte were house-foundations.

Regarding the function of the caps, wooden fenders similar in shape to Marianan latte tasa are used on house-piles by the Ifugao to prevent ingress of rats (2), and rat protections on posts are frequent in Oceania. It is probable that natural coral heads were originally placed upon the stone uprights in the Marianas for this purpose. The form evidently became stylized and was reproduced in stone where coral heads were not available, as for instance in the interior of Guam.


Turning to the historical sources we find the following significant data concerning this problem. Houses built on stone piles in Guam were actually reported by the Legaspi expedition in 1565, forty-four years after the discovery of the Marianas by Magellan. According to Burney (8) “Gaspar and Grijalva both describe the houses of these people (the Chamorros), that they were lofty, neatly built and well divided into apartments; the whole raised a story from the earth, and supported upon strong pillars of stone Fundadas sobre fuertes pilares de piedra. Besides these dwelling houses, they had others for their canoes, built like-wise with great stone pillars, one of which, near the watering place, contained four of their largest canoes.”

Comparing this statement with the archaeological data, we find no discrepancy. The uprights of latte could have functioned as piles to raise the house above the ground. Moreover, latte built on the shore with long axis perpendicular to it could obviously have served as canoe-sheds.

No other first-hand report of stone pillared houses in the Marianas has been found, but significant data concerning early Chamorro houses is given in several early accounts. Pigafetta (12), Magellan's historian, stated: “Their (the Chamorros') houses are all built of wood covered with planks and thatched with leaves of the fig-tree (i.e., banana - 460 tree) two brazas long; and they have floors and windows. The rooms and the beds are all furnished with the most beautiful palm-leaf mats …”

Sanvitores (7, Apr. 1937, p. 21), the Jesuit who lived in the Marianas as head of the first permanent mission from 1668 until he was martyred in 1672, reported: “The houses are the cleanest that have yet been found among Indios; built of coconut and maria (Calophyllum inophyllum) wood with roofs of coconut leaves curiously woven. They have four rooms, with doors or curtains of the same matting. One serves as sleeping room, another for storing goods, one as kitchen and the fourth is large enough in which to build and store boats.” Furthermore he added that the Chamorros buried their dead beneath certain of their houses (7, July, 1937, p. 10) and that property was ceremonially destroyed upon the death of a chief or noble (7, May, 1937, p. 20), significant statements in the light of the skeletal material found in and near latte. From the same source (7, April, 1937, p. 21) we learn that the natives lived in villages near their garden-lands along the coast and in hamlets in the fertile valleys of the interior, a distribution of population that corresponds to that of the latte.

When Anson visited Tinian in 1742, the latte were in ruins, but the natives told him that they had been the “foundations of particular buildings set apart for those Indians only who had engaged in some religious vow” (1). Anson's statement, repeated by later writers (3, 16, p. 45), that the stone structures were made with mortar is incorrect. The most spectacular ruins at Tinian (pl. 2) have long been known by the Chamorros as the “House of Taga” (4, 14, p. 18) and latte in general as casas de los antiguos (houses of the ancients; 6, p. 315).

Thus the historical as well as the archaeological evidence points to the inference that the latte functioned as house-sites. Turning to the problem of housing, we find that the Marianas islands, according to all the early reports, had a large population before the Spanish conquest. Sanvitores (7, April, 1937, p. 20) estimated the population of Guam alone to have been around 50,000 at the time of the first permanent mission. Even allowing these figures a wide margin of error, we must conclude with Tischner (16, p. 45)

- i
Latte site in ruinous state.
- ii
Latte ruin at Tiniau: “The House of Taga.”
- 461

that the number of latte ruins reported from Guam could not have housed the entire population. The latte must have been used for special types of houses only, such as men's club houses, reported in the Marianas by Sanvitores (7, April, 1937, p. 39), stone houses and canoe-sheds, reported by Gaspar, while the majority of the inhabitants lived in houses of wood as they do today (p. 8).

Men's houses disappeared from the Marianas with the introduction of Christianity because the Jesuits disapproved of the premarital consorting of the sexes institutionalized in them. The large canoes and also the canoe-sheds of the group vanished with the suppression of warfare and the spread of western influence.


The evidence presented above indicates that houses built on latte served at least two functions, namely as special dwelling and store-houses raised on stone pillars, and as canoe-sheds. There is practically no information available, however, concerning the structure of the houses which we assume were built on latte.

Freycinet (6, pp. 315-316; 5, pl. 81) has described and illustrated two types of stone houses reconstructed from information furnished him at Guam in 1819. However, since Freycinet visited Guam 150 years after the advent of the missionaries, it is unlikely that correct information concerning technical details was available. In one type, labelled the house of a powerful chief, a gabled roof which somewhat resembles that of a modern Chamorro dwelling-house described below, is supported by ten capped stone pillars of the type found at Taga, Tinian; in the other, labelled an ordinary house, a similar type of roof is supported by eight uncapped stone pillars.

The principal criticism of Freycinet's reconstruction is that they do not check with the historical sources quoted above. Moreover, although uprights without associated caps are seen in Guam today, whether or not they were originally capless is highly questionable.

It is possible, however, that the houses of the modern Chamorros may give a clue as to construction of ancient - 462 types. The most common dwelling-house now in use is a rectangular gabled type on piles of hard wood. The piles serve to raise the floor off the ground, not to support the roof as shown by Freycinet. In this respect their function is similar to the house-platforms or earth-mounds found sporadically, according to Tischner (16, map 3), in the Carolines, eastern Melanesia, and Polynesia. The walls are constructed of planks, split bamboo, or thatch of nipa (Nypa fructicans introduced from the Philippines, 13), pandanus or nete (Xiphagrostis floridula). Occasionally sheet-iron is used. The wooden piles are built in two parallel rows usually of four piles each, according to a ground-plan of the same form and dimensions as that of many of the small latte. Moreover, the piles are called haligi, the term used by the modern natives for the uprights in latte. The application of the Chamorro term meaning house-posts to latte uprights may, of course, be recent, since the natives regard the latte as house sites. On the other hand, if the latte did serve this function, its use in this connection is almost certainly ancient. The floor-space of the modern house is divided by a wooden partition into two compartments, a living-room and a bedroom. An outhouse attached to the rear serves as a kitchen.

It is possible that this house is a survival of an ancient type of dwelling. Eight capped uprights may have been used occasionally instead of wooden piles for unusually fine, well-built houses, probably as a sign of social prestige, which was exceedingly important among the ancient Chamorros.

The modern Marianan house type was probably intrusive from the west, either in prehistoric or historic times. Rectangular gabled houses of similar type, raised on wooden piles, are used as dwellings in many parts of the Philippines. Moreover, among the Bagabo and other Filipino groups, increase in size of houses is achieved by increasing the length, while the width remains constant as in latte-sites (10, p. 102). The use of stone pillars as house-foundations, however, is absent in the Philippines (10, p. 98), but house-burial is common.

In the Pelews southwest of Guam, however, we find a house-type built upon stone pillars remarkably similar in ground-plan and height to the twelve pillared latte of the - 463 Marianas. The type is described and illustrated by Kubary (11, pp. 221-225, pl. 29; figs. 5, 5a, 6, 6a). In the simplest type of Pelewan club-house, called keldok-bay, uprights of basalt are placed in two parallel rows, six to a row, according to a ground-plan similar in form and dimensions to the Marianan twelve-pillared latte. Moreover, the uprights are propped with small stones and built up to a horizontal plane to support cross-beams of wood. The walls are composed of wooden posts and plaited split bamboo. Six wide, window-like openings, used as entrances, have been left open, one in each end-wall and two in each side-wall. The high steep gabled roof of saddle-form is thatched with pandanus. Other club-house types in the western Carolines are built up on platforms of stone, not stone pillars. It is probable that the ancient twelve-stone pillared houses of the Marianas resembled the Pelew keldok-bay in structure as well as function.

The Pelewan canoe-shed (dianl) for housing large war-canoes, which has been described and illustrated by the same authority (11, pp. 265-266; pl. 48, figs. 5, 6) was a rectangular shed at least 8 metres long and 4 metres wide, the dimensions of a war-canoe. It had a high gabled roof of the usual Pelewan saddle-form, thatched with coconut-leaves and supported by eight unusually substantial posts of hard wood, 75 cm. thick and 1.40 metres high. Thus the shed was built according to a ground-plan similar in form and dimensions to the eight-pillared latte of the Marianas. It is probable that the stone-pillared canoe-sheds of the Marianas resembled these canoe-sheds of the Pelews.


Based on the archaeological, historical, and ethnographical evidence presented above, we may infer that the latte of the Marianas functioned as house-sites. Moreover, there were evidently at least two types of houses built on stone pillars in the Marianas, namely pile-houses and canoe-sheds. The details of construction of these house-types, however, are by no means clear.

Concerning the pile-houses, we know that the roofs were thatched. The floors were raised above the ground and the floor-space was divided into compartments which - 464 probably served as sleeping rooms, kitchens, storage, carpentry shops, etc. The walls of some houses at least had openings which served as windows. The ground under the houses was used for burials, accompanied by broken artifacts. We may safely assume that at least some of the pile-houses served as men's clubs. It is probable that club-houses resembling the Pelewan keldok-bay type were built on twelve-pillared latte, while houses resembling the modern Chammorran pile dwellings were built on smaller latte.

Concerning canoe-sheds, we may definitely infer that at least some latte built along the shore with long axis at right angles to the shore line served this purpose. Moreover, it is probable that canoe-sheds in the Marianas resembled sheds built for housing war-canoes in the Pelews, except that the thatched roof was supported by capped stone uprights instead of wooden posts.

  • 1. ANSON, GEORGE—A Voyage Round the World. London, 1748, p. 312.
  • 2. BATON, R. F.—“Ifugao lau,” University of California pub. in American Archaeology and Ethnology, vol. 15. 1919, pl. 17.
  • 3. CHRISTIAN, F. W.—The Caroline Islands. London, 1899, p. 54.
  • 4. CORTE Y RUANO CALDERON, FELIPE DE LA—Memoria descriptiva é histórica de las islas Marianas, Madrid, 1875, p. 84.
  • 5. FREYCINET, LOUIS DE—Atlas historique. Paris, 1825.
  • 6. FREYCINET, LOUIS DE—Voyage autour du monde. Paris, 1829, vol. 2, pt. 3.
  • 7. GARCIA, FRANCISCO—Vida y martirio de el venerable Padre Diego Luis de Sanvitores. Madrid, 1683. Translated in the Guam Recorder.
  • 8. BURNEY, JAMES—A Chronological History of the Discoveries in the South Seas or Pacific Ocean. London, 1803, vol. 1, p. 258.
  • 9. HORNBOSTEL, H. G.—Unpublished Field Notes on the Marianas islands, Bishop Museum, Honolulu.
  • 10. KROEBER, A. L.—Peoples of the Philippines. New York, 1928.
  • 11. KUBARY, J. S.—Ethnographische Beiträge zur Kenntnis des Karolinen Archipels.Leiden, 1895.
  • 12. PIGAFETTA, F. A., in BLAIR, E. H. and ROBERTSON, J. A., ed.—The Philippine islands,1493-1803. Cleveland, 1903-09, vol. 33, p. 99.
  • 13. SAFFORD, W. E.—“Useful Plants of Guam.” Contributions from the United States National Herbarium, vol. 9, 1905, p. 334.
  • 14. THOMPSON, LAURA—“Archaeology of the Marianas Islands.” Bishop Museum Bull. 100. Honolulu, 1932.
- 465
  • 15. THOMPSON, LAURA—The Native Culture of the Marianas islands (in preparation).
  • 16. TISCHNER—Die Verbreitung der Hausformen in Ozeanien. Quakenbrüch, 1934.
(Courtesy of the Bishop Museum)

Photographs by Hans Hornbostel.

  • 1. Latte at Hinapsau, Guam, consisting of eight uprights composed of limestone. One upright and all caps composed of coral heads have been displaced.
  • 2. Latte at Taga, Tinian, called “House of Taga.” The two capped limestone uprights, part of a series of twelve, are 16 feet high.
(Courtesy of the Bishop Museum)

Figures drawn by Gertrude Hornbostel except figure 1 which was drawn by Thomas Gill.

  • 1. Sketch map of Guam showing approximate distribution and density of latte, according to H. G. Hornbostel.
  • 2. Reconstructed type latte with eight capped limestone uprights, according to H. G. Hornbostel.
  • 3. A latte with twelve capped uprights at Meppo, Guam; a. plan, b. details of capped uprights number 7.
  • 4. Plan of a type latte with ten uprights, oriented with long axis parallel to the shore line.
  • 5. Plans of type latte oriented with long axis parallel to a river.
  • 6. Plan of a type latte with ten uprights, oriented with long axis at right angles to the shore.
  • 7. Cross section of a reconstructed latte site showing two capped uprights set deeply in the ground and reinforced below the surface with stone props; (1) extended burial with miscellaneous bones, (2) fireplace, (3) skull, (4) extended burial.
  • 8. Limestone upright with cap replaced, from a latte at Oruno, Guam; a. front facing the sea, west; b. side facing south; c. capital, circumference and base cavity; d. cross section of capital showing cavity. (From 14, fig. 6).
  • 9. Plan of a latte with eight uprights excavated by H. G. Hornbostel, showing position of skeletal material and artifacts revealed by excavations.
1   This study is part of the results of an ethnological field-trip to Guam in 1938-39, under the auspices of the University of Hawaii, the International Council of the Institute of Pacific Relations, and the Social Science Research Council. I wish to thank Dr. Felix M. Keesing and Mr. Kenneth P. Emory for criticism of the manuscript.