Granite is a coarse-grained intrusive igneous rock, typically composed of 20% - 60% quartz by volume together with alkali and plagioclase feldspar and smaller amounts of mica and accessory minerals ([5]). Its interlocking crystalline fabric gives it high compressive strength (commonly on the order of 100–250 MPa) and strong resistance to chemical and mechanical weathering compared with most sedimentary rocks, which is why granite landscapes tend to preserve joint-controlled and exfoliation-related landforms—domes, tors, sheeted faces—over very long timescales ([17]; [20]). These same properties explain why granite is difficult to quarry and shape with simple tools, and why worked granite is treated in the archaeological literature as a marker of investment and skill wherever it occurs.

Quartz crystals are piezoelectric: mechanical stress applied along particular crystallographic axes generates a measurable electrical charge ([2]). Because granite typically contains a substantial quartz fraction, this property has motivated a body of geophysical and archaeophysical research into whether granite outcrops or granite monuments produce macroscopic electrical or resonance effects. We are careful here to keep two claims separate. It is correct, and well established, that individual quartz crystals are piezoelectric. It is a distinct and considerably less settled question whether a granite mass—an assemblage of randomly oriented quartz grains bonded with feldspar and mica—produces a coherent, measurable piezoelectric effect at the scale of an outcrop or a wall, and whether any such effect had archaeological or cultural significance. We treat that second question as open and do not rely on it anywhere in the analysis that follows.

Across very different cultures and periods, granite is a repeated choice for monumental and megalithic construction, from the Nuragic towers of Sardinia to the dry-stone walls of Great Zimbabwe to the transported capstones of the Dutch hunebedden. Plausible reasons include exceptional durability and weathering resistance, high compressive strength, local availability in granitic terrain, and the symbolic or prestige value of working a famously hard stone. We present these as candidate explanations to be weighed against the specific evidence at each site, not as an established or complete list, and we do not treat granite’s durability or hardness as evidence, by itself, that any particular granite feature is anthropogenic.

Three Rivers lies in the western foothills of the Sierra Nevada, Tulare County, California (approximate town center: 36.4386˚N, 118.8971˚W), at the boundary between the granitic batholith terrain of the range and the alluvial Central Valley. The Sierra Nevada batholith is composed predominantly of Mesozoic granitic rocks emplaced during subduction-related magmatism, subsequently exposed by uplift and erosion; the region is well known for granite domes, tors, and exfoliation features (e.g., in Yosemite and Sequoia-Kings Canyon National Parks, immediately to the north and east). Kirch Flat Campground (36.8791˚N, 119.1501˚W), approximately 54 km (33.5 miles) north-northwest of Three Rivers, along the Kings River, lies within the same granitic terrain and is treated here as a second, geologically comparable locality. Precise coordinates for the individual documented features within the Three Rivers site have not yet been recorded and are a priority for the field survey proposed in Section 10; a preliminary, non-georeferenced inventory of the individual features illustrated in Section 7 is provided there. The broader region also has a substantial and well-documented archaeological record of Native American occupation extending deep into California prehistory ([10]).

Granite features previously documented in Montana include polygonal granite walls, dolmen-like arrangements of large slabs, balancing rocks, and standing monoliths at several independent localities, together with field measurements and orientation data collected during prior site visits. A full descriptive dataset comparable in detail to Section 7 below is being assembled and will be presented in a companion paper; here the Montana material is referenced only as comparative context. Because that feature-level dataset is not yet available in a form that supports a reproducible score, Montana is not scored against the Section 6 typology in the comparative table in Section 8.4; it is retained there, and here, as qualitative comparative material only.

The typology above is qualitative: a type is either present at a site or it is not. As a first step toward a quantitative treatment, we assign each type a provisional weight reflecting, approximately, how much co-ordinated effort or how improbable a natural weathering outcome it represents relative to the others—a single planar face is common and requires little explanation either way, while a lintel structure requires two or more separate blocks to be positioned in a specific load-bearing relationship. Table 1 lists the provisional weight assigned to each type. Summing the weights of the types present at a site gives a Morphological Complexity Index (MCI), reported for every site in the comparative table in Section 8.4.

Table 1.Provisional weights used to compute the Morphological Complexity Index (MCI) for each type (Section 6.1).

These weights are provisional and are offered as a starting point for discussion, not as a validated metric: they are not derived from a probability model of natural joint-controlled weathering, and a defensible version would need to be calibrated against the natural-control sites in Section 8.2, ideally using the quantitative joint-frequency and orientation data proposed for Phase 2 of Section 10. The MCI is reported here only to make the qualitative comparison in Section 8.4 slightly more informative; it is not treated anywhere in this paper as evidence of anthropogenic origin, and a high score at Three Rivers or Montana carries the same evidentiary weight as a high score at Yosemite or Joshua Tree—none, until it is tested.

Table 2.Proposed operational thresholds for prospective classification of Types A-H (Section 6.2). Values are provisional pending calibration against the natural-control sites in Section 8.2.

The Section 6 definitions (“close to a right angle,” “roughly horizontal rows,” “roughly planar levels”) are adequate for a first, descriptive pass but are not, on their own, reproducible between independent observers. To improve reproducibility, Table 2 sets out operational, provisional thresholds intended for prospective field application—that is, for use once the georeferenced measurement and photogrammetric survey proposed in Section 10, Phase 1, is under way. These thresholds are not applied retroactively to the existing photographic dataset in Section 7 (see Section 5); the Type assignments reported there remain qualitative. They are offered here so that future classification, by this author or by independent observers (Section 6.1; Section 10, Phase 1), can be checked against a stated, falsifiable standard rather than a verbal description alone.

At the Three Rivers locality, several rock exposures present as laterally continuous, roughly horizontal courses of rectangular to sub-rectangular blocks, separated by planar, near-vertical joint surfaces. Individual courses are traceable for several metres along strike. The exposure is shown in Figure 1, with a closer view of the joint planarity and course boundaries in Figure 2 (feature TR-01, Table 3).

Figure 1. Coursed granite exposure showing laterally continuous, roughly rectangular blocks separated by planar joint surfaces, Three Rivers, California. Photograph: Gary Bruce.

Figure 2.Close-range view of the same exposure showing joint planarity and course boundaries. Photograph: Gary Bruce.

Several localities show two or more large blocks in direct lateral or vertical contact, with contact surfaces that are close to planar. In one example, two rectangular blocks stand side by side, separated by a narrow, near-parallel-sided gap (Figure 3; feature TR-02); in another, a rounded-edge block and an elongated rectangular block rest adjacent to one another on a common base (Figure 4; feature TR-03).

Figure 3. Two adjacent rectangular blocks separated by a narrow, near-parallel-sided joint, Three Rivers, California. Photograph: Gary Bruce.

Figure 4.A rounded-edge block and an elongated rectangular block resting adjacent to one another, Three Rivers, California. Photograph: Gary Bruce.

The most visually striking feature documented is a stepped arrangement of large blocks rising in at least two levels, with a raised, roughly planar platform at one end resembling a stair tread (Figure 5, shown with a person for scale; feature TR-04). The feature is partially obscured by vegetation and by an overlying modern concrete water flume (Section 7.6). A closer view of the raised platform itself is given in Figure 6.

Figure 5. Stepped granite structure at Three Rivers, California, with a person for scale. Photograph: Gary Bruce.

Figure 6. Raised, roughly planar stone platform at one end of the stepped structure, resembling a stair tread. Photograph: Gary Bruce.

Away from the main stepped structure, the hillside above the Three Rivers campsite exposes numerous additional blocks and boulders, some standing as isolated angular monoliths (Figure 7; feature TR-05) and others forming discontinuous wall-like or terrace-like alignments over a broader area (Figure 8; feature TR-06).

Figure 7. Isolated standing angular granite block on the hillside above the main structure, Three Rivers, California. Photograph: Gary Bruce.

Figure 8.Discontinuous stepped alignment of blocks on the hillside, Three Rivers, California. Photograph: Gary Bruce.

An additional coursed exposure on the same hillside further illustrates the broader distribution of this morphology (Figure 9; feature TR-07).

Figure 9.Additional coursed exposure on the hillside, illustrating the broader distribution of similar morphologies (feature TR-07). Photograph: Gary Bruce.

Similar stepped and terrace-like granite morphologies were documented approximately 54 km (33.5 miles) north-northwest of Three Rivers, at Kirch Flat Campground along the Kings River, within the same regional granitic terrain (Figure 10; feature KF-01).

Figure 10.Stepped terrace-like granite feature at Kirch Flat Campground, California, a locality approximately 54 km (33.5 mi) north-northwest of Three Rivers. Photograph: Gary Bruce.

Two contextual observations bear directly on interpretation and are reported here rather than in the discussion, because they are themselves field observations. First, a nineteenth- to twentieth-century mining-era concrete water flume runs directly above the stepped structure at Three Rivers; portions of the flume are damaged and leak, and water runs across parts of the stone feature below. Second, a granite staircase adjacent to a house of early twentieth-century construction was documented at a separate location (Figure 11; feature TR-08); the homeowner reported uncovering it beneath vegetation, and its worked appearance differs in character from the surrounding informal landscaping.

Figure 11.Granite staircase adjacent to an early twentieth-century house, documented as site context rather than as a component of the main structure. Photograph: Gary Bruce.

These two observations matter for different reasons. The flume shows that at least part of the modern landscape immediately overlying the stepped structure is demonstrably anthropogenic and recent, which is relevant to any future discussion of surface exposure and possible erosional unroofing of the granite features below. The staircase shows that the boundary between clearly historic-era construction and the undated granite features is not always visually obvious in this landscape, which argues for caution in judging age or origin from appearance alone, in either direction.

8.1.1. Nuraghi, Menhirs, and Tomb of Giants—Sardinia, Italy (Type D, F, G)

The comparative approach adopted here builds on earlier anthropological work examining recurring technological and engineering characteristics of megalithic construction across geographically and culturally distinct sites ([12]). Sardinia’s Nuragic megalithic tradition (Bronze Age, c. 18th-8th century BCE) provides an unambiguous anthropogenic granite reference: coursed dry-stone towers (nuraghi) with corbelled interior vaulting, single standing monoliths (menhirs), and chambered gallery graves (Tombe dei Giganti, “Tombs of the Giants”). These photographs were taken during a field visit to Sardinia in November 2020, illustrated in Figures 12-15.

Figure 12.Coursed and stacked granite boulders at a Nuragic site, Sardinia, Italy.

Figure 13.Standing granite menhir, Sardinia, Italy, shown for scale.

Figure 14.Entrance stele of a Tomb of the Giants (Tomba dei Giganti), a Nuragic chambered gallery grave, Sardinia, Italy.

Figure 15.Interior corbelled dry-stone construction inside a nuraghe, Sardinia, Italy, showing coursed masonry technique.

The nuraghi are useful comparanda specifically because the construction technique—coursed, roughly rectangular stone laid in horizontal courses with corbelling toward the interior—is independently documented and undisputed ([8]).

8.1.2. Great Zimbabwe (Type D)

Great Zimbabwe, in Masvingo Province, Zimbabwe (20˚16'S, 30˚56'E), is a UNESCO World Heritage Site built and occupied between the 11th and 15th centuries CE by ancestors of the Shona people ([15]; [3]). Its dry-stone granite walls, some reaching 11 m in height, are coursed without mortar and include the largest single ancient structure in sub-Saharan Africa south of the Sahara (Figures 16-18).

Great Zimbabwe is the strongest single comparandum in this paper for coursed dry-stone granite construction at landscape scale: the walls are unambiguously anthropogenic, the coursing technique is well documented, and the construction date is independently established.

Figure 16.Great Zimbabwe, Masvingo Province, Zimbabwe: coursed dry-stone granite wall of the Great Enclosure, with a person for scale.

Figure 17. Aerial view of the Great Enclosure at Great Zimbabwe, showing the curved coursed granite wall circuit.

Figure 18.Narrow passage between two parallel coursed granite walls, Great Zimbabwe.

8.1.3. Hunebedden—Drenthe, Netherlands (Type G, H)

The Netherlands preserves 54 hunebedden (megalithic chambered tombs), 52 in Friesland/Drenthe and 2 in Groningen, built by the Funnel Beaker Culture more than 5000 years ago from granite glacial erratics carried south from Scandinavia by ice-age glaciers ([1]). Structurally each hunebed is a dolmen: smaller upright support stones (orthostats) carry larger, near-horizontal capstones. Hunebed D27, near Borger (52˚55'48''N, 6°47'51''E), is the largest in the country at 22.6 m long, with 26 supporting stones; it was first excavated in 1685 by Titia Brongersma. Hunebed D50, near Noord-Sleen (52˚47'46''N, 6˚47'14''E), retains a near-complete ring of 24 kerbstones around its passage (Figures 19-21).

The hunebedden raise a logistical question directly analogous to one a future study must ask about any anthropogenic component at Three Rivers or Kirch Flat: not only “is the block shaped”, but “how, physically, could it have been moved and placed” by the society credited with it.

Figure 19.Hunebed D50, Noord-Sleen, Drenthe, Netherlands: granite capstones resting on upright support stones, with a person for scale.

Figure 20.Detail of capstone-to-orthostat contact, Hunebed D50, Noord-Sleen, Netherlands.

Figure 21. General view of a hunebed passage grave, Drenthe, Netherlands, showing the alternating pattern of upright and horizontal granite blocks.

8.1.4. Osirion, Abydos, Egypt (Quarrying, Transport, and Engineering Context)

The Osirion, a subterranean structure behind the Temple of Seti I at Abydos, is built from red granite quarried at Aswan, roughly 300 km to the south, together with quartzite ([7]). Its date is debated: it is conventionally associated with the adjoining temple, but its megalithic style and construction technique differ noticeably from typical New Kingdom architecture, and that association has been questioned. We take no position on the dating question here; it is not relevant to why the site is cited in this paper. Individual blocks are reported at up to roughly 60 tons; the photograph below, taken during a site visit, shows blocks in this range (Figure 22). Systematic survey of the Aswan granite quarries themselves—including the tool marks, extraction scars, and unfinished blocks left in situ—has been used elsewhere to distinguish ancient quarrying activity from natural rock breakdown ([6]), a distinction directly relevant to the H1 prediction in Section 3.

Figure 22.Massive granite blocks in the Osirion, Abydos, Egypt, reported at up to roughly 60 tons.

Osirion is referenced here for its relevance to quarrying, long-distance transport, and construction logistics rather than to landscape morphology or chronology: it documents that granite blocks in the tens-of-tons range were quarried, moved roughly 300 km, and precisely fitted, which sets a concrete point of comparison for what is achievable without the lifting and transport equipment available today, independent of when the work was carried out.

8.1.5. Mahabalipuram, India (Type C, F—Contested Boundary between Natural Outcrop and Carving)

Mahabalipuram (Tamil Nadu, India) preserves seventh- and eighth-century CE monolithic rock-cut shrines and reliefs carved directly from granite outcrops, alongside unfinished and uncarved granite boulders in the same landscape. It is a useful comparandum precisely because it documents, at a single site, the full transition from unmodified granite outcrop to finished monument, which is the transition California and Montana observations are trying to evaluate (Figure 23).

8.1.6. Giza, Egypt (Accepted Anthropogenic Construction, Partly Granite)

Giza is included for completeness because it is frequently invoked in discussions of this kind, but it is a limited comparandum for landscape-scale granite morphology: the bulk of the visible structure is limestone, and granite use is confined to specific transported elements ([7]). Its greater relevance is to construction engineering and quarrying technique (Section 8.1.4) rather than to the block-and-terrace morphology documented at Three Rivers and Kirch Flat (Figure 24).

Figure 23.The Shore Temple, Mahabalipuram, Tamil Nadu, India: a granite structure carved and assembled directly on a coastal granite outcrop.

Figure 24.The Giza pyramid complex, Egypt (29.9792˚N, 31.1342˚E). Granite is confined to specific internal elements (e.g., the King’s Chamber and relieving chambers of the Great Pyramid) and to elements quarried at Aswan; the visible superstructure is primarily limestone.

8.1.7. Akapana, Tiwanaku, Bolivia (Accepted Anthropogenic Construction, Partly Granite)

The Akapana pyramid at Tiwanaku, Bolivia, part of a broader ten-site comparative survey of pyramid engineering ([14]) and of the independently documented Tiwanaku monumental complex ([21]), incorporates granite blocks alongside sandstone and andesite in its terraced construction. The photographs below, taken during a site visit, show fitted rectangular blocks with recessed joint-fittings characteristic of Tiwanaku masonry (Figure 25).

Figure 25.Akapana pyramid, Tiwanaku, Bolivia. (a) Fitted rectangular granite block with recessed joint-fittings. (b) A dressed rectangular block on the terrace surface. (c) Site signage showing the reconstructed stepped-pyramid plan. (d) Fitted blocks at a terrace corner.

Akapana is listed here as a further accepted anthropogenic comparandum for terraced (Type E) and coursed (Type D) granite-inclusive construction; a fuller treatment with site-specific morphological typing is a candidate for a future revision.

The following localities are widely documented in the geomorphological literature as natural granite landforms, without independent archaeological evidence of human modification ([17]; [9]). They are included to define the baseline against which any claim of anthropogenic origin at Three Rivers, Kirch Flat, or the Montana localities must be tested (Figure 26).

Yosemite National Park, California (Type A, E)—granite domes and sheeted exfoliation faces produced by unloading and jointing ([17]).Joshua Tree National Park, California (Type B, C)—blocky granite jointing and boulder piles from subsurface weathering along orthogonal joint sets.City of Rocks National Reserve, Idaho (Type A, B, F)—spires, domes, and orthogonal-jointed outcrops in the Almo pluton, a further Sierra-adjacent example of joint-controlled granite morphology at landscape scale.Dartmoor, England (Type E)—granite tors formed by differential weathering along joint-bounded corestones, frequently cited as the natural-process baseline for stepped granite outcrops ([18]).Matobo Hills, Zimbabwe (Type H)—extensive granite balancing rocks, conventionally attributed to spheroidal weathering and subsequent removal of surrounding grus.Huangshan, China (Type F)—natural granite peaks and pillars produced by joint-controlled erosion, frequently used as a visual comparandum for standing-monolith morphology ([19]).

A note on Matobo Hills specifically: the balancing-rock morphology there is treated in mainstream geomorphology as a settled natural-process case, and it is listed in that column in Section 8.4 on that basis. The author does not consider that classification closed for every individual formation at the site and regards at least some of them as open to discussion in the same sense as the California and Montana material; that view is personal, is not argued for in this paper, and does not change how Matobo Hills is used here—as the field’s own natural-process reference case for Type H.

Figure 26. Natural granite reference landscapes. (a) Yosemite National Park, California—granite domes and sheeted exfoliation faces ([17]). (b) Joshua Tree National Park, California—blocky granite jointing and boulder piles. (c) City of Rocks National Reserve, Idaho—jointed granite spires and outcrops of the Almo pluton. (d) Dartmoor, England—granite tors formed by differential weathering. (e) Matobo Hills, Zimbabwe—granite balancing rocks, conventionally attributed to spheroidal weathering; see note below. (f) Huangshan, China—granite peaks and pillars from joint-controlled erosion.

Three Rivers and Kirch Flat, California (Section 7), and the Montana localities (Section 4.2) are the two questioned sites evaluated in this paper. Neither is classified here as natural or anthropogenic; both are treated, on typological grounds alone, as candidates for the systematic investigation proposed in Section 10.

Table 4 below scores each site for the presence (✓) of each morphological type defined in Section 6, whether the site has independent archaeological evidence of human construction, and its provisional Morphological Complexity Index (MCI, Section 6.1). This is a qualitative first pass based on the material presented here and is intended to be replaced by quantitative feature counts once the survey work in Section 10 is complete; the MCI in particular is not evidence of origin (Section 6.1) and is included only as a compact summary of how many, and which, types co-occur at each site. Montana is not scored in this table: the feature-level dataset needed to support a reproducible score is being assembled for the companion paper referenced in Section 4.2, and Montana is accordingly retained here as qualitative comparative material only.

Table 4.Comparative summary of morphological types, independent archaeological status, and provisional MCI across the questioned, anthropogenic-control, and natural-control sites (Section 8.4).

Types: A = planar faces; B = orthogonal intersections; C = polygonal block arrangements; D = coursed walls; E = stepped terraces; F = standing monoliths; G = lintel structures; H = balancing megaliths. MCI = Morphological Complexity Index (Section 6.1), a provisional, non-evidentiary weighted sum. *See Section 8.2 for the author’s personal reservation about the Matobo Hills classification, which does not change how the site is used in this table.

A separate but related body of field documentation concerns near-spherical stone artifacts, some of them granite. These are a distinct morphological phenomenon from the eight types defined in Section 6 and are not treated as evidence for or against the origin of the Three Rivers or Kirch Flat material; they are noted here because several examples are specifically granite, as opposed to the volcanic, sandstone, or limestone spheres documented at other sites, and therefore fall within the material scope of this paper (Figures 27-30).

Figure 27.Polished granite sphere near the Hagia Sophia, Istanbul, Turkey.

Figure 28.Granite sphere near Šibenik, Croatia.

Figure 29.Granite sphere, Teočak, north-eastern Bosnia and Herzegovina.

The best-known stone spheres in the world, from the Diquís Delta region of Costa Rica, are carved from granodiorite rather than granite proper. Granodiorite is a coarse-grained intrusive igneous rock closely related to granite, distinguished mineralogically by a higher proportion of plagioclase relative to alkali feldspar; the two are often discussed together informally, but we keep the distinction explicit here for the same reason we do so elsewhere in this paper (Section 2.1). Over 300 Diquís spheres are documented, some exceeding 2 m in diameter.

Figure 30.Granodiorite spheres, Diquís Delta region, Costa Rica. (a) In situ, partially overgrown. (b) With a person for scale, showing near-perfect sphericity. (c) A museum display of spheres of varying diameters. (d) A sphere mounted for outdoor public display.

Spherical shaping raises a different question from the coursed and stepped morphologies documented in Section 7: a sphere is a much less probable outcome of joint-controlled weathering than a planar face or a right-angle intersection, which makes the natural-versus-anthropogenic question comparatively easier to test case by case (via petrography, internal fabric, and comparison with known spheroidal weathering products) even though it remains open at most individual sites. A fuller treatment of the granite and granodiorite examples in this catalogue, alongside the volcanic, sandstone, and limestone spheres documented elsewhere, is given in a separate study ([13]).

The present study is limited in ways that bear directly on how its conclusions should be read, and we set them out explicitly rather than leaving them implicit.

Incomplete exposure—most of the documented features are partially buried, and their full extent below the current ground surface is unknown. Sections currently interpreted as isolated blocks or short wall segments may connect to larger structures, or may not; excavation is the only way to resolve this (Section 10, Phase 4).Vegetation cover—dense moss, leaf litter, and brush obscure joint surfaces and contact relationships at Three Rivers in particular, which limits how confidently individual joint planes can be traced or measured from photographs alone.Absence of excavation—nothing has been excavated at either the California or Montana localities. Every observation in Section 7 is a surface observation; subsurface stratigraphy, buried artifacts, and construction fill (if any) remain entirely unknown.Absence of LiDAR—no digital elevation model or LiDAR survey exists for either locality, so feature extent, precise dimensions, and site-wide spatial relationships cannot yet be measured quantitatively; the typology in Section 6 was applied by eye, from photographs.Absence of GIS—without georeferenced feature locations, the spatial-clustering and alignment questions raised in Section 9 cannot be tested statistically; the comparative table in Section 8.4 is a qualitative first pass for exactly this reason.Absence of geophysical survey—no ground-penetrating radar, resistivity, or magnetometry data exist for either locality, so the subsurface continuity of any feature, and any distinction between surface-exposed natural jointing and a buried anthropogenic structure, remains untested.Absence of independent inter-observer classification—the Type A-H assignments in Section 7 and Table 3 were made by a single observer. No second qualified observer has yet classified the same imagery independently, so no inter-rater reliability statistic (e.g., Cohen’s κ) can yet be reported. This is formalized as a required task of Phase 1 in Section 10.

None of these limitations is fatal to the paper’s argument, because the paper’s argument does not depend on them: the claim is that the documented morphology justifies the investigation in Section 10, not that the investigation has already been carried out. But they do mean that every interpretive statement in this paper—about either the natural or the anthropogenic baseline—should be read as provisional pending exactly the data these limitations describe.