The Delhi Iron Pillar Does Rust. The Real Explanation Is Better Than the Myth.

The buried section of the Delhi iron pillar is covered in rust. When the Archaeological Survey of India excavated its footing in 1961, the underground surface carried rust layers running from a few millimetres to fifteen, with numerous cavities and corrosion pits across it. That is not an awkward detail for a footnote. It is the fact that makes everything above ground explicable.

The exposed shaft has corroded too, only very slowly. Bardgett and Stanners measured its surface film with a permanent magnet gauge and found roughly 200 micrometres, a fifth of a millimetre. Gosta Wranglen, assuming parabolic growth and using corrosion data for mild steel in Delhi's air, had predicted almost exactly that for sixteen centuries. A film that thickens as the square root of time is throttling its own growth, which is what protective means.

R. Balasubramaniam of IIT Kanpur established what that film is. The iron carries about a quarter of one per cent phosphorus, high by modern standards, and under repeated wetting and drying that phosphorus builds a thin, dense, crystalline phosphate layer pressed against the metal. The mechanism appeared in Corrosion Science in 2000 and remains the standard account. It is more interesting than the mystery it displaced, because what the Gupta smiths achieved was a process rather than a secret ingredient.

A pillar, an inscription, and an argument about where it used to stand

Seven point two one metres of solid wrought iron, 1.12 metres of that buried and a little over six above ground, roughly forty one centimetres across at the base, weighing more than six tonnes, with a decorative bell capital on top. It has stood in the courtyard of the Quwwat ul Islam mosque at Mehrauli for some eight centuries.

The oldest inscription runs to three Sanskrit verses in the shardulavikridita metre, cut in the eastern variety of the Gupta script. Bhau Daji Lad published the first reading that correctly gave the king's name as Chandra, in the Journal of the Bombay Branch of the Royal Asiatic Society in 1875. John Faithfull Fleet produced the critical edition in 1888.

"Having in faith fixed his mind upon (the god) Vishnu, this lofty standard of the divine Vishnu was set up on the hill (called) Vishnupada."

Closing verse of the Chandra inscription, in J. F. Fleet's translation, Corpus Inscriptionum Indicarum, volume III, 1888. The parentheses are Fleet's own.

The earlier verses praise Chandra for battles in the Vanga country and against the Vahlikas beyond the seven mouths of the Sindhu. Govardhan Rai Sharma dated the script in 1945 to the first half of the fifth century, on its resemblance to the Bilsad inscription of 415 and to the Baigram and Kahaum inscriptions of the middle decades of that century. Balasubramaniam, assembling numismatic, archaeological and literary evidence, identified Chandra as Chandragupta II Vikramaditya.

Where Vishnupada was, and who moved the pillar to Delhi, is not settled. Fleet said Mathura, which later work rejected. Meera Dass and Balasubramaniam argued in 2004 for Udayagiri, where Chandragupta II left dated cave inscriptions, and Michael Willis extended that case in 2009. Finbarr Barry Flood argued the same year for Vidisha. Alexander Cunningham credited the Tomara king Anangpal. One object, four answers, no agreement.

What the rust is actually made of

Take the metal first. Averaging the published analyses, Wranglen put it at roughly 0.15 per cent carbon, 0.25 per cent phosphorus, 0.005 per cent sulphur, 0.05 per cent manganese, traces of silicon, copper and nickel, the rest iron. Individual figures scatter, because the pillar is not uniform: Hadfield found 0.114 per cent phosphorus in 1912, Ghosh 0.28 in 1963. The popular figure of one per cent is wrong. A quarter of one per cent is the number that matters, and in modern steelmaking that much is a defect, because it leaves steel liable to crack during hot working.

Balasubramaniam sampled rust from just below the bell capital, a region no visitor can reach and therefore the oldest undisturbed rust on the object. X-ray diffraction, infrared spectroscopy and Mossbauer spectroscopy gave a clear answer. The scale is mostly iron oxides and oxyhydroxides in amorphous form, plus something crystalline that ordinary rust does not contain: iron hydrogen phosphate hydrate.

The sequence he reconstructed runs like this. The iron is full of microscopic slag particles and specks of unreduced ore, and those carry carbon from the smelting charcoal, which makes them cathodic with respect to the nearly pure iron around them. Each sets up a tiny galvanic cell, so the pillar begins by corroding faster than plain iron would. That early corrosion concentrates phosphorus at the boundary between metal and scale: Ghosh measured 0.35 per cent phosphorus in the rust against 0.18 in the metal beneath.

Phosphorus at that interface catalyses a compact layer of an amorphous oxyhydroxide called misawite, the same phase that protects modern weathering steels and that forms only in patches on ordinary mild steel. Then the slow part. Each time the surface wets and dries, phosphoric acid forms there, iron phosphates precipitate, and atmospheric oxygen drives the iron in them from the plus two state to the plus three until they reorganise into crystalline iron hydrogen phosphate hydrate. That reorganisation collapses the porosity of the layer, and moisture and oxygen can no longer reach the metal.

The pillar is not protected against rust. It is protected by its rust. The film had to grow before it could defend, which is why the early centuries were the dangerous ones.

The older environmental explanation was incomplete rather than wrong. Delhi's relative humidity does not stay above seventy per cent for long stretches of the year, and W. H. J. Vernon's experiments showed that iron rusts little below that threshold.

Climate alone cannot be the answer, though, because comparable high phosphorus ancient Indian iron has survived in places nothing like dry: the iron beams in the Surya temple at Konarak on the humid Odisha coast, the iron pillar at the Mookambika temple at Kollur in the Kodachadri hills. Material and environment are both required. Neither is sufficient. And the protection is specific to air: immersed in dilute salt or sulphur dioxide solution, samples of pillar iron corroded briskly. Submerged, this metal would be gone.

The third of the pillar that nobody photographs

The stone platform around the base did not exist before the middle of the nineteenth century. J. D. Beglar built it when he re-erected the pillar, and laid a coating of lead, about three millimetres thick and uneven, over the buried surface. The intention was protective.

Lead is cathodic with respect to iron. Coupling the two in wet soil does not shield the iron, it drives the iron's corrosion wherever the coating has a gap, and Beglar's coating had gaps. The soil around it was later found loaded with soluble sulphates and chlorides, which makes the couple work harder. When the ASI dug the footing out in 1961, the lead was in excellent condition. The iron underneath was pitted and carrying up to fifteen millimetres of scale.

The rust down there is chemically similar to the rust above, with differences that follow from the setting. Oxygen in soil is restricted, and under restricted oxygen the dominant product is magnetite, which is what Ghosh's diffraction work on the magnetic fraction found, alongside lepidocrocite and goethite. Nor is there any wetting and drying cycle to force the crystalline reorganisation, so Balasubramaniam infers the phosphate stayed amorphous and porous. Same chemistry, denied the one input it depends on, and it does not protect.

The part of the pillar exposed to Delhi's weather has looked after itself for sixteen centuries. The part deliberately protected in the nineteenth century has been damaged by the protection. A new lead coating went on when the pillar was refounded in the 1960s, against the wishes of the Survey's own Chief Chemist, and Balasubramaniam argued in print for replacing it.

Getting it right without knowing why it works

The phosphorus was not a mystery additive. It is a consequence of how the iron was smelted.

Indian smiths of this period reduced high grade ore in the solid state using charcoal, never melting the metal, then hammered the spongy lumps to squeeze out slag. Modern blast furnace practice adds lime. These furnaces did not, which excavated Indian ironmaking slags confirm: they are fayalitic, iron silicate, with no calcium oxide. Phosphorus leaves metal for slag efficiently when lime is there to receive it and poorly when it is not. No lime means phosphorus stays in the iron. That is not a decision. It is what falls out of the recipe.

There is a hint of intent elsewhere in the tradition, and it should be labelled a hint. Francis Buchanan, describing eighteenth century steelmaking in Karnataka, recorded a furnace at Devaraya Durga where clay crucibles were charged with wrought iron and bark from Cassia auriculata, which draws high phosphorus from the ground. That is thirteen centuries after the pillar and cannot be read backwards onto it. It does suggest the effect of some plant materials on iron had been noticed somewhere and repeated.

The fabrication tells the same story. The pillar was built by forge welding bloomery lumps of twenty to forty kilograms, heated to around 700 to 900 degrees and hammered onto the growing shaft, most probably with the shaft lying horizontal. The joins held. When somebody fired a cannonball at it from close range, about four metres up, the ball dented the metal and cracked it horizontally on the far side without breaking the shaft. The gunner was either Nadir Shah's army in 1739 or Ghulam Qadir's in 1787, and the sources do not settle which.

Finishing was done by hammering, which left the outer skin as ferrite depleted of carbon, and phosphorus migrates into regions where carbon has been depleted. So an operation carried out for appearance raised the phosphorus concentration in exactly the few millimetres of metal where the protective film would later form. Balasubramaniam builds that chain across several steps rather than one, and it is the least direct link in his case.

No one in the fifth century could have known any of this. Galvanic cells, passivation and oxidation states are nineteenth and twentieth century ideas. What the smiths had was a reliable procedure and long observation of which procedures produced lasting iron.

Wootz, and what actually reached Damascus

The same tradition produced the crucible steel that Persian and Arab smiths turned into the most sought after blades of the medieval world. The English word wootz comes from Dravidian: urukku in Tamil and Malayalam, ukku in Kannada and Telugu, from a root meaning to melt.

The process sealed wrought iron and carbonaceous plant matter, bamboo and leaves among them, into clay crucibles and fired them long and hard, in the region of 1200 to 1400 degrees, until the iron took up carbon and melted. The product was hypereutectoid, well over one per cent carbon, beyond what a bloomery can reach. Sharada Srinivasan has recovered steel prills of roughly one to one and a half per cent carbon from crucible remains at Mel-siruvalur in Tamil Nadu, and crucible fragments are known from Kodumanal, Konasamudram and Gatihosahalli.

Now the part that gets garbled. What went west was the ingot, not the blade. Wootz cakes were traded to Persia and the Arab world and forged into swords there, and the weapons took their name from Damascus, a forging and trading centre rather than a source of steel. The watered pattern comes from bands of clustered iron carbide particles that form in a hypereutectoid ingot under repeated forging and thermal cycling. John Verhoeven and Alfred Pendray, who gave the first reproducible modern account of the banding in the late 1990s, traced it to carbide forming trace elements inherited from the ore, vanadium in particular, at a few thousandths of one per cent. European pattern welded steel, confusingly also called Damascus, gets its figure by forge welding strips of different alloys. Two unrelated techniques, one name.

Michael Faraday analysed wootz in 1819 and blamed its qualities on alumina and silica. He was wrong.

Zawar, and a metal that will not stay put

The zinc site at Zawar near Udaipur belongs to the medieval period, not the ancient one, and the distinction is worth insisting on. Mining there is genuinely old, with radiocarbon dates running back to roughly the fourth century before the Common Era. Distillation is a separate and later matter: the earliest dates from the retort heaps fall around the twelfth century, industrial working begins about then, and it runs until the works were abandoned in the early nineteenth. Conflating mining dates with smelting dates is the commonest error made about Zawar, and it is unnecessary, because the medieval achievement is remarkable on its own terms.

Zinc is hard to win for a reason copper and iron do not share. It boils at 907 degrees, and reducing zinc oxide with carbon needs temperatures above that. The metal leaves the furnace as vapour, and vapour meeting air, or the furnace's own carbon dioxide, turns straight back into zinc oxide. You cannot tap zinc from the bottom of a hearth. You have to catch the vapour where no air reaches it and chill it fast.

Zawar's answer was distillation downward: banks of furnaces holding rows of inverted clay retorts charged with ore, the vapour travelling down a tube into condensers below the fire. Paul Craddock and colleagues at the British Museum, working with Indian collaborators from 1982, excavated the sites and found retorts in the tens of thousands, which is what an industrial process looks like on the ground. Europe did not distil zinc commercially until William Champion patented his method at Bristol in 1738.

A Sanskrit text describes the process. The Rasaratnasamuccaya, attributed to Vagbhata and generally placed in the thirteenth or fourteenth century, sets out tiryakpatana, distillation by descent, with the condensing vessel beneath. Text and excavated furnace describe the same operation, and agreement that direct between a Sanskrit technical treatise and an industrial site deserves to be cited more often.

Two men called Nagarjuna, and a book that was never written

Here the tradition has confused itself badly enough that the confusion is part of the record.

Nagarjuna the philosopher, founder of the Madhyamaka school, lived around 150 to 250 of the Common Era. Nagarjuna the alchemist is a considerably later figure, of disputed date, whose name attaches to a rasashastra work called the Rasendramangala. Indian tradition merged the two, and modern popular writing merges them harder, so the author of the Mulamadhyamakakarika is regularly credited with recipes for treating mercury.

The tangle goes further than mistaken identity. Prafulla Chandra Ray, in his pioneering A History of Hindu Chemistry, whose two volumes appeared in 1902 and 1909, described a work called the Rasaratnakara by Nagarjuna, and on his authority it entered the bibliographies and stayed. Dominik Wujastyk went back to the manuscripts and found Ray had worked from a codex in the Raghunatha Temple Library at Jammu that physically mixes leaves of two books: the Rasendramangala, which is Nagarjuna's, and the Rasaratnakara, which the manuscripts uniformly ascribe to Nityanatha Siddha. There is no Rasaratnakara by Nagarjuna. There never was.

Ray was not careless. He was a serious chemist working on damaged and unedited sources, and he made a reading error anyone handling a composite manuscript could make. The correction arrived eight decades later, in Ambix in 1984, in a paper Wujastyk titled "An Alchemical Ghost". It was a book with no text, cited by people who had not read it, because there was nothing to read.