HistoricalChristian.Faith

Do the "lathe-turned" pillars of the medieval Deccan imply machining beyond their period?

Type: supporting argument (off-timeline) · Assessment: weak — as the claim that the turned pillars of the Hoysala temples, the Kopeshwar temple at Khidrapur and the Vijayanagara stonework at Hampi are the residue of a lost technology surviving from a much older age. The two halves fail in different places and are rated separately. The manufacturing claim fails because the lathe is not the exotic part: turning soft stone on a rotating spindle is an attested ancient craft — Pliny describes it, and Tamil artisans at Namakkal still do it by hand today — and the Hoysala stone is chloritic schist, soft enough at the quarry to cut with an iron chisel. The dating claim fails harder still, because at each of the three sites the anchor is contemporary writing: Belur carries its own foundation inscription of 1117, Somanathapura a dated stele of 1258, Khidrapur about a dozen inscriptions naming its kings, and Vijayanagara was described in operation by a Persian ambassador in 1443 and by two Portuguese chroniclers who walked its streets. The genuine puzzle underneath — how precise the turning actually is — has never been measured in any published study, which is worth saying plainly in both directions · advanced by alternative-antiquity documentarians and ancient-technology writers

Related arguments: out-of-place artifacts (dates under skepticism) — see also whether Kailasa at Ellora predates its eighth-century date · Liangzhu's jades and the technology they required · whether this reconstruction's witnesses are securely dated · the vimanas and divine weapons of the Sanskrit epics · the Ta Prohm "stegosaurus" carving · Angkor's engineering and the antiquity of the Khmer capital · predynastic Egypt's hard-stone vases · the Great Sphinx's water-erosion redating · the Osireion at Abydos · the Giza radar scans and the "second Sphinx" · Sacsayhuamán's polygonal walls

See also: the Watchers as teachers of the arts (assessment: contested), the mechanism an inheritance reading of this argument would need.

The claim

Three sets of objects are usually put together, and they do share a single visual signature: rotational symmetry in stone, on a scale and to a finish that looks industrial. The Hoysala temples of Karnataka — Chennakeshava at Belur, Hoysaleshwara at Halebidu, Keshava at Somanathapura — stand on porches of cylindrical, bell-bodied, deeply fluted pillars whose ring-grooves run true all the way round; the technique is so characteristic that the standard architectural literature simply calls them "lathe-turned," and one description of the construction has it that "lathe-turned cylindrical pillars are a distinctive feature of the Hoysala architecture."¹ The Kopeshwar Mahadev temple at Khidrapur, on the Krishna at the Maharashtra–Karnataka border, carries the same signature in its open Swarga Mandapa, a circular hall whose forty-eight pillars stand in three concentric rings.² And at Hampi, the ruined Vijayanagara capital, two loose stones lying near the Mahanavami Dibba — a flat slab bearing concentric circular grooves around a central socket, and a cylinder that fits it — are read as the surviving bed-plate and spindle of a vertical lathe.

The argument built on this runs in two steps. First the manufacturing step: rotational symmetry at this precision, in stone, cannot be reached by hand. In its strongest published form the claim is quantified — "surface variations of less than 0.5mm across entire pillar sections, indicating a level of control that rivals modern precision engineering," a twelve-foot pillar with sixty-four decorative bands spaced exactly 2.25 inches apart, and modern masons with period-appropriate tools failing to match the originals. Then the dating step: if the technique exceeds the medieval toolkit, the objects are not medieval, and the inscriptions that date them record a later patron rather than the original maker. The Hampi stones are pressed hardest here. "These perfect circular marks on this rock can be produced only with machines," runs the claim; "they are perfectly concentric and there is no deviation, no chipping," and "the rectangular slot in the center proves that this rock was machined using lathe technology" — from which, "this set up actually proves that ancient people used vertical lathe technology, at least 500 years ago," with the further suggestion that the site itself may be far older than the empire credited with it.³ On this reconstruction's premise, a technique that outruns its culture is a survival — something carried down from a world before the Flood.

The official dates, and the methods behind them

The method at all three sites is epigraphic and, at Vijayanagara, documentary — inscriptions cut on the buildings themselves and, in the Hampi case, named foreign visitors who saw the city standing. Nothing radiometric is involved anywhere, which should be said openly; but as anchors go these are the strong kind rather than the weak kind, and the difference from a monument dated by style alone is the whole point.

Hoysala. Chennakeshava at Belur was commissioned by Vishnuvardhana in 1117, and the temple says so: an inscription on the wall of the main mandapa records the commission and the dedication to Vijayanarayana, and the image itself carries an inscription on its pedestal saying the king caused it to be made. The complex holds more than a hundred further inscriptions running from 1117 into the eighteenth century.⁴ Keshava at Somanathapura is anchored the same way and more tightly: a stele inside the Mahadvara pavilion, in Old Kannada, records that Somanatha, a general under Narasimha III, obtained his king's permission to build the temple, established the agrahara, and endowed its maintenance — dated 1258 (a minority reading gives 1268).⁵ These are foundation records on the object, not later grants filed elsewhere.

Kopeshwar. Attribution runs to the Shilahara kings of Kolhapur, Gandaraditya and his successors, across roughly 1109–1178, with later Yadava work. The temple carries about a dozen inscriptions; "all these inscriptions except one are in Kannada language and script," the exception being a Sanskrit text in Devanagari on the outer wall by the south entrance, recording the Yadava Singhana II — the inscription that also records his defeat and imprisonment of the Shilahara Bhoj II, which ended the Kolhapur line. Here a real caveat belongs: popular accounts of the site do circulate a seventh-century origin with the twelfth-century work as renovation, and the temple's building history plainly has phases. That caveat is entered on the argument's side, not against it.⁶

Vijayanagara. The city was founded in 1336, sacked in 1565 and never reoccupied, and it is one of the best-witnessed pre-modern cities in Asia. Abd al-Razzaq Samarqandi, ambassador of Shah Rukh, was in it from April to December 1443 and described its concentric fortifications and the fields and gardens between them. Domingo Paes saw it about 1520 and Fernão Nuniz about 1535; both chronicles survive and were translated by Robert Sewell in 1900. Behind them lies the imperial epigraphic record and, at the site, the ASI's own museum catalogue.⁷

What would have to hold true for a much older date

Three conditions. Each has genuine precedent somewhere, and each is conceded here at full strength.

(a) Rotational symmetry at this precision is unreachable by hand-rotation and abrasion. This is the load-bearing premise, and it is the one the whole argument stands on. If a hand-driven rig cannot produce these surfaces, something else did, and everything downstream follows.

(b) The Hampi stones are machine components. A slab with concentric turning grooves and a central socket, plus a cylinder that seats in it, is a suggestive pair; if the identification as bed-plate and spindle holds, then a purpose-built machine tool existed at the site, and the burden shifts to explaining what else it could be.

(c) The inscriptions and the eyewitnesses record renovation or occupation, not origin. The precedent is real and is documented elsewhere on this wiki: kings appropriating predecessors' monuments is ordinary practice, from the sphinx of Amenemhet III re-inscribed by three later kings to the widespread surcharging of statuary under Ramesses II; and the formula "the king built" routinely covers rebuilding, enlarging or re-consecrating something already standing.⁸ At Khidrapur, where phased construction is actually attested, this is not a hypothetical.

Adversarial assessment: do the conditions hold here?

The lathe is not the anomaly. Turned stone is an attested ancient craft, and it is still practised. Condition (a) fails first and hardest, and not on modern authority but on ancient testimony. Pliny, writing in the first century and drawing on Theophrastus three centuries earlier, records it flatly: "In the island of Siphnos there is a stone that is hollowed out and turned on the lathe so as to form cooking utensils or tableware; and this I myself know to be the case also with the green stone of Comum in Italy." The stone he means is steatite — soapstone. Turning soft stone on a lathe was a known Mediterranean trade a thousand years before the Hoysalas, in the same material, and the technique was unremarkable enough that Pliny mentions it in passing while cataloguing soft stones.⁹ Nor did it lapse. In Namakkal in Tamil Nadu artisans still shape blocks of local soapstone on manual lathes, without modern machinery, into the thick-walled kalchatti cooking vessels used across the southern states — the same operation, the same material, in a living workshop that no one calls mysterious.¹⁰

The material was chosen because it is soft. Hoysala temples are built of chloritic schist — soapstone, a talc-schist, at the very bottom of the hardness scale, and softer still when freshly quarried, hardening only on exposure to air. That property is the standard explanation for why Hoysala carving is deeper and finer than the Chalukyan sandstone and Chola granite work around it: the sculptors were cutting something closer to hard soap than to granite. The construction accounts describe exactly the process the objects imply — a block roughly prepared, "then turned on a lathe" and "finally carved into various shapes and sizes using an iron chisel," with the turning done about a vertical axis on a heavy swivel and the chisel held against the moving surface, chisel marks still visible on the finished pillars.¹¹

The pillars are not monoliths, which is what makes the turning tractable. A fourteen-foot Hoysala pillar is not one piece of stone spun on a bearing. It is a stack of squat cylindrical drums, dry-set and pinned with iron rods dropped into matching sockets — mortarless construction relying on weight and interlock. Each drum is a manageable workpiece of a size a rope-and-animal drive can turn, and the descriptions of the technique say plainly what drove them: "the prime movers for these lathes were horses and elephants." The same tradition is confident enough with rotation to build it into the finished building: the Narasimha pillar at Belur was mounted to revolve on its own axis, and "between the pillar and the walls of the receptacle once rolled stone ball bearings, and that's how the pillars rotated."¹² A culture that puts a stone thrust bearing under a temple pillar for show is not a culture struggling to conceive of a turntable.

The precision figures have no published basis. This cuts both ways and both should be said. The sub-millimetre surface variation, the 2.25-inch band spacing and the micrometre-level 3D scan attributed to the Indian Institute of Science do not trace to any study that can be found; they circulate on ancient-technology sites without a citation, and nothing in the peer-reviewed literature on Hoysala architecture reports a metrology survey of the pillars. So the quantified version of the claim is not evidence — but neither is there a published measurement showing the tolerances are loose. What can be said is that the qualitative observation is genuine and is not in dispute: the symmetry is excellent, and it is excellent because turning about a fixed axis makes every groove a circle concentric with that axis automatically. Rotational accuracy is what a lathe gives you for free; it is the property of the method, not a measure of the toolmaker's sophistication. That is why the argument's central inference — extraordinary roundness therefore extraordinary machinery — does not follow. A hand-cranked rig and a powered one produce the same circles; they differ in how fast, not in how round.

The Hampi stones read as an unfinished workpiece, and nobody has identified them otherwise. Condition (b) is weaker than it looks. The slab does show, as its most careful visitor-description puts it, "turning marks.. like in a lathe but the rotational axis is vertical" around a centre socket — and the same observer, with no stake in the mainstream account, concluded the piece "seems like a work-in-progress job survived and somehow damaged," possibly meant to be a wheel. That reading fits the site: Hampi is littered with abandoned stonework, and its quarry faces and half-split boulders still carry the diagnostic Vijayanagara method — a line of holes three or four inches deep at four- to six-inch intervals, dry wooden pegs driven in and soaked so that the swelling wood splits the granite along the line.¹³ And the identification as machine parts has never been supported from the object: no accession record, no measured drawing, no wear pattern analysis, no companion pieces. Against that, the ASI's own museum at Kamalapura catalogues actual lathe-turned objects from the site and labels them so — a schist pillar of the Vijayanagara period from behind the Guards' House, and a second schist pillar assigned to the pre-Vijayanagara period, bell-shaped and deeply fluted, its carved Ranganatha with Sridevi and Bhudevi showing the hand-work laid over the turned form.¹⁴ The excavators are not concealing a lathe. They wrote "lathe" on the label.

Usurpation has nothing to work on at Hampi, and little at Belur or Somanathapura. Condition (c) is the one that transfers worst. Re-inscribing a monument moves a name; it does not move a city that a Persian ambassador walked through in 1443 and two Portuguese chroniclers described as a functioning capital in the 1520s and 1530s. To push Vijayanagara's stonework back before the Flood is not to reinterpret an inscription but to discard three independent eyewitness accounts of the place in use, plus the imperial epigraphy, plus a documented sack in 1565. At Belur and Somanathapura the mechanism is only slightly better placed: these are foundation records cut on the buildings themselves, naming the patron, the deity, the endowment and the year — not a third party's later report of somebody else's temple. The one site where the renovation reading has real purchase is Khidrapur, whose phased history is genuinely attested; and there it buys centuries at the outside, in a region whose whole temple sequence is dated around it, not the millennia the argument needs.

The technique has an ancestry and a descent. The Hoysala lathe was not invented at Belur and did not die there. The sculptors took it up under Chalukya influence — Chalukyan turned pillars are bolder and plainer, the Hoysala ones more intricate — and the ASI's own pre-Vijayanagara turned pillar at Hampi sits between the two traditions chronologically. A lost antediluvian technology does not show up as a regional craft with visible predecessors, visible refinement, visible successors, and a modern descendant still turning cooking pots in Tamil Nadu.

The manufacturing sub-claim, kept separate

Suppose the dates were untouched. One could still hold that the pillars are beyond hand production and that whoever made them, whenever they worked, had tools we do not know about. Rated on its own, that claim is weak too, and for reasons that do not depend on any date.

The requirements are modest and each is met independently. A turning axis: attested in the same culture as a finished architectural feature, on stone ball bearings, at Belur. A drive: rope and animal power, which the Deccan had in quantity and which the construction accounts name. A workpiece of manageable size: supplied by the drum-and-dowel construction, which cuts a fourteen-foot pillar into pieces a metre or so tall. A soft enough material: chloritic schist, freshly quarried, is about as workable as stone gets. A cutting tool: an iron chisel held against the rotating face, the marks of which survive on the pillars. And a proof of concept in the same material: soapstone vessels turned on manual lathes, attested by Pliny in the first century and still made in Tamil Nadu now. Nothing in that list is missing, exotic, or inferred.

What is genuinely impressive is not the machine. It is the deployment: an eleventh-to-thirteenth-century Deccan state running enough turning rigs, over enough decades, to produce hundreds of pillars no two of which repeat the same carving — a labour and organisational achievement, which is a different thing from an out-of-place technology.

The inheritance reading, and why no Flood horizon rescues it

There is a deeper form of the manufacturing claim than either sub-claim above, and it deserves to be stated at its strongest because it does not depend on moving any date. It grants the medieval dates and holds instead that the capacity was inherited — that turning stone was antediluvian know-how, among the arts the Watchers of 1 Enoch 7–8 are said to have taught, carried through the Flood into a post-Flood heir. On coherence the reading has one bar to clear first: the maker must have survived the deluge. It clears it easily, and that should be conceded. The Hoysalas worked in the twelfth and thirteenth centuries, the Shilaharas and Yadavas across the twelfth and thirteenth, Vijayanagara from the fourteenth to the sixteenth — every one of them some three millennia after even the latest biblical Flood. So where the same reading elsewhere leans on the Septuagint long chronology (Flood c. 3300–3000 BC) to leave a straddling maker-culture standing after the water — as it must at Liangzhu — here the horizon choice does no such work: on the Masoretic c. 2348 BC or the Septuagint reckoning alike, the Deccan makers are heirs and not casualties. The Flood date, load-bearing at straddling sites, is simply not load-bearing here.¹⁵

But coherence was never where the reading was going to fail. An inheritance argument needs work to do — a capacity gap that transmitted antediluvian craft would fill — and turned stone leaves none, because the craft is documented from end to end on the post-Flood side of both horizons. It has an ancestry: the Hoysala sculptors took the lathe up under Chalukya influence, and the ASI's own museum at Hampi catalogues a pre-Vijayanagara turned pillar sitting chronologically between the two traditions.¹⁴ It has a refinement, a set of successors, and a living descendant — artisans still turning soapstone cooking pots on manual lathes at Namakkal today, the same operation in the same soft mineral Pliny watched turned on the lathe in the first century.⁹¹⁰ Every requirement the finished pillars imply — a turning axis, a rope-and-animal drive, a manageable drum-sized workpiece, a soft chloritic-schist blank, an iron chisel held to the moving face — is attested in the Deccan itself, none of it exotic or inferred.¹² And there is not even the straddle pattern an inheritance story elsewhere can appeal to — a technique attested before the Flood and climaxing after it — because the whole lineage, Chalukyan predecessors included, is post-Flood: there is no antediluvian phase to inherit from at all. Whatever an heir could have carried through the water is at most ordinary skilled stone-turning, which the Deccan demonstrably worked out and handed down in full view. The Flood date thus decides nothing here: the inheritance reading is coherent and empty, and the manufacturing claim stays weak on evidence no choice of horizon disturbs.

Assessment

Weak — as the claim that these works are significantly older than their dates and reflect a lost or pre-Flood technology; and weak separately as the claim that their manufacture exceeds the attested medieval toolkit. The deepest form of the claim — that the turning capacity was antediluvian know-how carried through the Flood, not a redated artifact — is coherent for so late a maker (the Deccan dynasties are post-Flood on every reckoning, heirs and not casualties) but comes out weak as well, because it needs a capacity gap to fill, and a craft with visible ancestry, living descendants and a first-century Mediterranean parallel leaves none.

The concessions owed are three, and none is trivial. The precision has never been measured in any published study, so the confident textbook sentence about hand-craft and the confident ancient-technology sentence about micrometre tolerances are both unsupported by data. The Kopeshwar temple's building history really is phased, and the seventh-century origin story that circulates about it is not absurd on its face — a renovation read as a foundation is exactly the failure mode this class of argument exists to catch. And the Hampi stones near the Mahanavami Dibba really do carry vertical-axis turning marks around a central socket, which is a curious object that no one has published on.

What defeats the reading is that its central inference is backwards. Rotational symmetry is not the hard part of stone turning; it is the part the method supplies automatically, the moment there is an axis. So finding excellent circles tells you a lathe was used and tells you nothing about how the lathe was powered or when. And the lathe itself is not out-of-place: Pliny watched soft stone turned on one, in the same mineral, in the first century, and the trade is alive in Tamil Nadu today. Once the manufacturing premise goes, the dating step has nothing pushing it, and it runs straight into anchors of the strongest kind available short of radiocarbon — foundation inscriptions cut on the buildings at Belur and Somanathapura, a dozen inscriptions at Khidrapur, and at Vijayanagara a Persian ambassador and two Portuguese chroniclers who describe the city while people were living in it. Appropriation can move a name on a monument. It cannot move a monument out from under an eyewitness.

The honest residue is a research question rather than a mystery: nobody has actually put a metrology survey on a Hoysala pillar, and nobody has published on the Hampi discs. Both would be worth doing, and either could sharpen this page. Neither is likely to move it, because what would have to be found is not a tighter tolerance — a hand-driven lathe can hold a fine one — but a tolerance no rotating rig could hold at all, which is not what a lathe argument can ever produce.

Sources