The South Oculus

Canterbury Cathedral

Brian Hall


  South transept exterior showing the south oculus in its architectural context
  The south transept of Canterbury Cathedral and its oculus window before conservation
  Close-up of the south oculus with its external iron framework

Metalwork is usually painted for a very good reason: paints and coatings both protect metal from corrosion and provide a decorative finish.

When selecting a protective coating for metals we are faced with a multitude of choices: paints based on synthetic or natural binders (such as alkyds, acrylics, vinyl-acrylics, vinyl acetate/ethylene (VAE), polyurethanes, polyesters, melamine resins, epoxies or oils), hot and cold zinc spray coatings, electroplating, and waxes.

In addition, surface preparation options for historic metalwork are also numerous and equally diverse, so deciding on a suitable treatment is never as straightforward as it may appear to be at first.

The south oculus window at Canterbury Cathedral is a case in point.

Oculi are large Romanesque circular windows that are divided into segments by iron armatures and not stone tracery as in the case of rose windows which replaced the oculi after the mid-13th century.

Canterbury Cathedral has two oculi, one in each of the two transepts, but only the south oculus retains its ironwork complete.

With a diameter of almost 4.5m the south oculus is an internationally important early ferramenta (iron framework) structure which is believed to date from between 1178 and 1180. Current research shows it to be unique in that it has a contemporary secondary external frame which offers a technological solution to supporting large ecclesiastical windows prior to the use of stone tracery.

The usual ferramenta which were used to support earlier, smaller, windows were inadequate on their own where such a large window was concerned, so additional support was provided by an in situ secondary, external iron grille, tied to the primary framework at regular intervals. The secondary grille performs the same function as a space frame in contemporary structures, providing support and rigidity while still keeping the structure relatively lightweight.

The ferramenta had survived for 800 years with little intervention and was in a stable condition. It was made entirely in charcoal iron which has a reputation for longevity and corrosion resistance.

A safety assessment was carried out on the structure to determine the effects of wind load using ambient vibration tests. Once the modal response had been identified, the structure could be digitally modelled using DIANA finite element analysis software. It was established that 17 per cent of the structure had been lost through corrosion and that it could afford to lose a further 75 per cent before it became structurally unstable.

The grille junctions consist of wrought iron horizontal bars passing through a square punched hole in the vertical bars. These joints were tightened with one or more copper wedges driven in, and then sealed with molten metallic lead and red lead putty to prevent water ingress.

The stained glass leaded panels are supported off the rear of the ferramenta by iron wedges and lugs riveted through the ferramenta. Again, putty was used to create a watertight seal.

When the South transept was scaffolded as part of a wider programme of repairs, the Dean and Chapter required an Icon-accredited metal conservator with suitable experience to advise the consultant team made up of the surveyor of the fabric, the head of stained glass conservation, the head of conservation and masonry, and the structural engineer. As the appointed metalwork consultant, I was to put forward suggestions/options for investigations, carry out treatment trials and provide a report showing treatment specification.

The client needed a long-term, low-maintenance solution, as once the scaffolding to the south transept was taken down, this part of the cathedral might not be easily accessible for another 50 years.


As its general condition was stable, the ferramenta was able to cope with the forces acting upon it. However, the north oculus external grille had already been lost, demonstrating that there was a potential for failure here too. Caution had to be exercised.

In the ferramenta’s recent history it had clearly been painted with several layers of a lead based paint, which was now flaking and peeling. The paint had not been consistently maintained, and had broken down, enabling water ingress and corrosion. Failed, broken down and degraded paint on ironwork promotes differential aeration corrosion, and can be more detrimental to substrate iron than no coating at all.

  Close up of the copper wedges
  Copper wedges in a joint showing how little of the material has been lost to corrosion despite being over 800 years old

On the internal, upper and lower faces of the ferramenta, in addition to the paint layers, there were the remains of old window putty, which had dried out and was now absorbing moisture. This moisture sitting against the iron surface for extended periods of time had resulted in small areas of extensive flaky corrosion.

Corrosion products covered the exposed surface of the iron and could be seen beneath the flaking paint. However, the rust streaking and staining of the stone appeared to be light considering the volume of exposed iron.

Lead-filled grille junctions were also a cause for concern, as dissimilar metals (such as lead, copper and iron) in direct and electrochemical contact could easily become sites for electrochemical (galvanic) corrosion. However, lead caulking and putty seemed to adequately protect the joints from water ingress, and air flow around these areas was slowing down the rate of corrosion. There was no discernible difference in the amount of corrosion at these junctions compared to other parts of the grille.

There were various old repairs and breaks to the grille and ferramenta, but they were not detrimental to the overall stability of the structure. Sections of metal pipe had been fixed around the horizontal bars in areas of breaks, but they were now loose, performing no structural function, and as they risked forming water traps they could lead to corrosion. Once these additions were removed, it was determined that underlying breaks had been stable for a long time. They did not compromise the structural integrity of the grille and did not require interventive treatment. Strap repairs on the ferramenta were left in place.


The conservation treatment to the metalwork was to be carried out in situ. A minimum intervention approach was adopted. All parts of the metalwork were to be conserved as found wherever possible. Interventions would be restricted to stabilising, and repairs would not be undertaken unless they were absolutely necessary. Where necessary, repairs were to be made with compatible ferrous materials to prevent galvanic corrosion. No effort was to be made to straighten distorted metal components.

We needed to decide on a coating system that would offer a low level of intervention, provide adequate protection against weathering, be easily maintainable or retreatable, and be compatible with the original materials.

The use of both traditional and modern paints was ruled out as our brief was to offer a treatment that could last up to 50 years with little maintenance. This led us to favour the trialling of oils and waxes.

We had to consider an adequate coating for both the external and internal faces of the ferramenta and grille, as well as a sealant for the interface between the ferramenta and the leaded stained glass panels. Compatibility between these materials was to be explored during the trials.


Very few archaeo-metallurgical studies have been carried out on comparative architectural ironwork in the United Kingdom. The conservation of the south oculus provided an opportunity to study the extensive iron components of the window to better understand its manufacture and the history of changes during its lifetime. Small samples were removed for metallographic analysis, and sections of both original and repair material were examined with a portable X-ray fluorescence (XRF) reader.

  A conservator uses a X-ray fluorescence reader to analyse
  Using a portable XRF analyser to document the chemical composition of the copper wedges as they age

Following investigation and cleaning trials, it was decided to adopt a low-intervention cleaning strategy: this would involve the removal of the paint, putty and corrosion products using scalpels, soft wire brushes and Wishab sponges. (Wishab sponges are dry cleaning sponges for use on soiled surfaces. Dirt particles are absorbed into the sponge’s surface which, in turn, crumbles away during use.)

Three different coatings were tested: tannic acid suspended in polyvinyl acetate (PVAc), microcrystalline wax, and Waxoyl, a hydrocarbon suspension of wax and phosphoric acid rust inhibitors made by Hammerite Products.

These coatings offer a good barrier layer between the ironwork and the moisture in the atmosphere and are relatively easy to maintain. Unlike flaking paint, they do not require complete removal before re-application. During the testing process it was found that Waxoyl gave the best result and appeared to be the hardest wearing.

During the cleaning tests a small amount of linseed oil putty was applied to the coated areas, with the purpose of assessing how well the putty would stick to the various coatings. The putty needed to provide a weather seal between the external environment, the wrought iron frame and the leaded glazing. As expected, the tests showed that the putty would not stick to the Waxoyl and therefore another approach had to be considered for the inside face of the ferramenta.

One solution was to treat internal surfaces with tannic acid, with linseed oil putty applied on top. Tannic acid is a well established corrosion inhibitor for iron. It works by transforming the destructive iron (III) oxides into iron tannates, which are stable compounds. However, the tannic acid is known to cause corrosion of lead. To ensure that the tannic acid did not affect the lead matrix of the stained glass panels, two layers of Paraloid B72 (20 per cent weight by volume in a solvent of acetone and industrial methylated spirits) were applied over the tannic acid. Linseed oil putty could then be safely applied to the sealed surface.

During the conservation process it was noted that a number of the glazing lugs were broken or extremely thin and delicate. It was agreed that 21 new lugs would need to be fabricated and fitted. However, the original material, high-quality charcoal wrought iron, is no longer available commercially, so alternatives had to be considered.

  Back of the open ferramenta after removal of the stained glass
  The back of the ferramenta with surviving lugs riveted through the iron bars to provide the connection point to support the stained glass
  Close-up showing repaired lug and corroded surface
  Detail of a past repair to secure one of the lugs: 21 of them had to be renewed entirely

The medieval bloomery technology of iron smelting, where the metal was never melted, leads to an inhomogeneous microstructure as found in the samples taken from the oculus. Variation was evident within the few millimetres of a single sample. This type of charcoal or bloomery iron smelted at low temperatures, is relatively pure, with only small amounts of carbon and phosphorus absorbed into the metal, the former from the fuel, and the latter from the ore.

One option was to make the lugs from modern rolled/recycled wrought iron. The recycled wrought iron available today is generally puddled iron from the industrial age. It contains slag inclusions and laminations, and has a similar grain structure to charcoal wrought iron. Charcoal-smelted bloomery iron is surprisingly resistant to corrosion in contrast to more recent iron and steel in which sulphur from the use of mineral fuels is retained within the metal and acts to accelerate the corrosion process.

The original samples examined were also very low in slag inclusions, which may serve as easy routes for corrosion penetration. Furthermore, due to the reclamation process modern wrought iron varies widely in quality and would be visually and metallographically indistinguishable from historic repairs, which might confuse future investigations.

The alternative option was to use modern pure iron. Metallurgically there is little difference between pure iron and wrought iron other than the fact that the pure iron does not contain fibrous slag inclusions. The conservation benefits of using pure iron are that it is close to wrought iron, of consistent quality, and easily distinguishable from the original material.

After considering the options the decision was made to use pure iron. A historic repair to one of the glazing lugs was used as the design basis for the new lugs. A small design change was incorporated into the new lugs, and all the new components were also date stamped to make them easily identifiable in the future.

Considering the slight chemical difference between the wrought and pure iron, it was important to minimise the risk of galvanic corrosion. To do this, at least one of the three corrosion causing factors – oxygen, water, or direct contact between dissimilar metals – had to be eliminated.

Using techniques that were originally used in the manufacture of the oculus (surrounding the copper wedges), lead paste (a compound of linseed oil and red lead containing, by weight, 95% lead (II, IV) oxides) was applied to the ferramenta to act as a jointing compound and barrier between the wrought and pure iron. The new lug was then placed over the paste and locked into place using iron wedges. The lead paste would act as a seal preventing water ingress.

This brightly coloured lead paste was allowed to harden for two weeks, and was then toned down with a coating of black Waxoyl.


  Detail of the oculus after conservation; the metalwork has been finished in a dark Waxoyl coating
  The high performance wax coating penetrates and protects the whole metal surface, and its distinctive lustre reveals the laminations of the historic material. It requires only relatively simple maintenance every five to seven years to ensure protection.

The south oculus ironwork had clearly undergone some deterioration: localised areas of more advanced corrosion were evident in places, and substantial repairs had been made to the ferramenta.

However, it is a tribute to the ingenuity and skills of the medieval craftsmen that the iron window framing had continued to fulfil its role for over 800 years. It even survived damage received during World War II bombing raids, when a section of the ferramenta was effectively removed under enormously high stress loadings, but the integrity of the frame held.

Having carefully considered our options, a high-performance wax coating (Waxoyl) was chosen rather than either a traditional or a modern paint system. The maintenance of this coating could be carried out every five to seven years, using a cherry-picker to gain access to the oculus. The ferramenta and grille would require a simple wash-down, followed by re-application of Waxoyl.

Paint systems, by comparison, generally require more invasive maintenance, including shot-blasting in the case of the two-pack systems commonly used, and all paint systems can trap moisture when they crack or flake.

We believe that high performance wax coatings provide a good balance of low-intervention treatment and manageable maintenance. It is hoped that this sympathetic approach to the conservation of its 800-year-old ironwork will preserve the oculus for many more centuries.


The conservation and repair works to the south oculus window were carried out as part of the ongoing conservation works to the cathedral’s stained glass windows and was assisted by a generous donation by the Worshipful Company of Ironmongers.



The Building Conservation Directory, 2014


BRIAN HALL ACR is managing director and a senior conservator at Hall Conservation Ltd. He has worked in the field of conservation for the past 28 years specialising in sculpture and architectural metalwork. He is also a trustee for the National Heritage Ironwork Group and tutor on the SPAB courses ‘The Repair of Old Buildings’ and ‘Metalwork Masterclass’.

Further information


Churches (general)

Stained glass



Stained glass
Site Map