Brighter than Right, Part 2: Technical Problems of Protection, Health and Safety at St Paul’s Cathedral

On June 15th the BBC news website reported that a £40m 15 years long restoration of St Paul’s cathedral by “state-of-the-art conservation techniques” had recovered Sir Christopher Wren’s “original vision” and left the building “as fresh as the day it was completed”. Major restorations invariably generate breathless accounts of recovered original glories made by vanquishing the “grime of centuries” – but grime only ever dates back to the previous restorations. At St Paul’s these were in the early and late twentieth century and the proposals for this last restoration explicitly declared that far from returning the interior to Wren’s original painted scheme, it would be stripped to a never-intended, never encountered state of bare-stone whiteness – see Part 1. As for the operation being “state-of-the-art”, consider these defensive/confessional remarks by David Odgers, of Nimbus Conservation, in 2005 just after the remains of Wren’s paint had been stripped: “Being completely inexperienced in the use of the material at the beginning, the learning curve was steep and problems of protection, health and safety issues and night time application had to be addressed”. This is the story of that learning curve.

The method used for cleaning the St Paul’s interior was novel and experimental but as such it was unproven. In both its composition and its effects it earned censure from leading conservation experts (see below). The cleaning agent was an adapted, commercially available, latex rubber poultice laced with a mix of chemicals that were said to comprise an agent tailored to be similar to the mild alkalinity of Portland stone – a special version of the “Arte Mundit” water-based paste manufactured by the Belgian company FTB Restoration. The instigator/director of the restoration, the architect and the 17th Surveyor to the Fabric at St Paul’s Cathedral, Martin Stancliffe, admitted (at a lecture on October 21st 2003) to having slim knowledge of matters chemical and of having devolved – “entrusted” – responsibility for the application of the new paste to the Nimbus conservators (who were learning on the job while the cathedral remained in full commercial and ecclesiastical use).

This state arose despite Mr Stancliffe’s boast that Nimbus had been selected as contractors after “the optimum formulation of the material had been achieved.” In practice, Nimbus, being entirely unfamiliar with this supposedly thoroughly researched and tested material, found its application by hand to be “slow, messy and to leave a streaky appearance on the cleaned stone.” Thus, when this multi-million pounds single-sponsor restoration was approved and underway it was discovered that: a) the result would look awful; b) it would take forever; and, c) it was leaving a terrible smell (of ammonia) throughout the cathedral.

With the restoration in full progress, the manufacturer went back to the drawing board and radically changed the paste’s approved composition and method of application. It has not been made clear of what the chemical changes consisted or whether approval for them was obtained (see below). With thousands of square metres of stonework to be cleaned, FTB Restoration devised what Mr Stancliffe and Mr Odgers described as “a method for spraying on the material using compressed air with a specially designed pump and nozzle”. This enabled each restorer to apply in “only a few minutes” up to 3.5kg of chemically laced latex paste per square metre (See Figs. 1,2,3,4 & 5). The industrial speed of application – between 50 and a 100 square metres a night – and, with it, the wider commercial prospect of buildings remaining open to business during interior restorations, caused great excitement in the upper tiers of heritage administration. With 2m visitors a year to St Paul’s and admission charges then at £6, now at £14.50, closing St Paul’s during the interior restoration would likely have lost something in the region of £50-60m, but, as we will see, the technical “solutions” to the initial unanticipated problems created serious consequences of their own.

Approval for the Arte Mundit cleaning method had been given by The Cathedrals Fabric Commision for England in November 1999 following claimed earlier approvals by a bevy of heritage watchdogs: English Heritage; SPAB – The Society for the Protection of Ancient Buildings; The Victorian Society; and The Georgian Group. It is not possible to establish the precise chemical basis on which formal approval was given by the CFCE because, in breach of good conservation practice, the three technical parts of the eight parts submission document have been withheld on grounds of commercial confidentiality. For information on technical matters we must rely on the cathedral’s own fluctuating (and self-contradicting) published accounts, on our correspondence with Mr Stancliffe (which was terminated by him in March 2003), and on documents obtained by cathedral employees whose health was affected by the restoration (see below).

In December 2002, Mr Stancliffe and Mr Odgers gave a joint account of the ongoing restoration in Conservation News. They explained why the Arte Mundit poultice method had been adopted and why the so-called “Mora Poultice” method had been rejected. It might be noted that the latter is a cocktail of thixotropic paste, sodium bicarbonate, ammonium bicarbonate, detergents and the aggressively powerful chelating agent EDTA – ethylene diamine tetra-acetic acid. That poultice had been designed for cleaning marble buildings and was used experimentally on Michelangelo’s Sistine Chapel ceiling frescoes to disastrous effect (see our post of April 1st). By 1992 research had shown that the brightness produced was not a product of marble surfaces having been cleaned but of their being etched by the EDTA into dissolved irregularities which scatter light in all directions.

It was thus known before this restoration began that consuming stone is a consequence of EDTA levels being either too high or left too long on the surface. The Mora poultice was also rejected because the copious amounts of water needed to remove it would have turned St Paul’s Portland stone brown when, as we saw in Part 1, Mr Stancliffe’s ambition was to produce white stripped-stonework in defiance of Wren’s original warmly toned oil painted decorative scheme.

Mr Stancliffe and Mr Odgers reiterated in their joint Conservation News article, that the chemical composition of their Arte Mundit paste had been “specifically formulated” after a great deal of research (but by then the research seemed to have run into the restoration itself). While it had been found necessary at the outset to add EDTA to the latex paste, they said, this had been done only “at a concentration of 2000mg/kg (0.2%)” precisely to avoid injuries to stone when used at solutions of 11% in the Mora poultices. Before discussing the hugely varying EDTA levels seen to have been used at St Paul’s, consideration should be given to Arte Mundit’s initial principle cleaning ingredient – the pungent alkali ammonia.

Sprayed applications compound the health risks associated with hazardous chemical products. Although Mr Stancliffe and Mr Odgers admitted “the downside of using compressed air is that the Arte Mundit is applied as a fine particulate and releases ammonia into the atmosphere”, they seemed to regard this as a nuisance rather than a threat to health. Until then, as they put it, the paste had “contained ammonia” but, because “St Paul’s is visited by thousands of people each day, it would be inappropriate for the Cathedral to smell of ammonia.” They added that “Recent developments have meant that the concentrations of ammonia have been significantly reduced in the Arte Mundit so that potential risk has been minimised.” It was not there said of what the “developments” consisted or to what figure the ammonia had been reduced, but it was admitted that atmospheric concentrations of ammonia (which is generally detectable at 4ppm) had reached 10ppm in the cathedral. Needless to say, reducing the smell of the most pungent chemical ingredient is not the same as eliminating or reducing the risks presented by all the chemicals present in the sprayed paste.

Sand blasting the interior surfaces had been rejected partly because installing absolutely airtight isolation to retain airborne dust would have been too expensive (see Figs. 4, 5, 6, 7, 8 & 9). In April 2003 a Health and Safety Executive officer (who was under the impression that the only change made to the Arte Mundit paste had been a reduction in its ammonia content and who seemed unaware that EDTA had been incorporated even though it was listed in the manufacturer’s safety data sheet to which she referred – see Figs. 10, 11 & 12) reported, in a seeming counsel of ineffectuality, that “In order to clean a large, old cathedral it is expected that dust will become airborne and it may be that this is contributing to the respiratory problems [of staff members]”. Despite not finding any trace of what she termed “EDA” anywhere in the cathedral, she had “asked the contractor to improve the ventilation in the work areas during the cleaning process.” (See Figs. 3, 4 & 5.) Quite mystifyingly, she added that this EDA, “was not a separate substance within the mixture but we wished to ensure that it was not produced as a by-product.”

When in October 2002 a member of the cathedral’s staff who had suffered severe skin afflictions requested a copy of the manufacturer’s material safety sheet (Figs. 10, 11 & 12), she discovered that the paste was then containing EDTA at up to10% – which is to say, almost the same as in the discredited (and rejected) Mora Poultice, and therefore at up to fifty times greater than the figure shortly to be claimed by Mr Stancliffe and Mr Odgers in the December Conservation News.

That safety sheet was no rogue document. In March 2003, when declining to answer our questions on inconsistencies in the official accounts, Mr Stancliffe produced certain “fact sheets” which, he averred, “answered all potential questions which you or anyone else may have on this aspect of our interior cleaning programme.” The sheets (entitled “The Arte Mundit Fact File”) were models of un-clarity and consisted entirely of questions jointly put to the restorers and to the Arte Mundit manufacturer by themselves, along with the answers they gave to their own questions. Thus, to “What controls are in place to ensure that the application and removal of the Arte Mundit is competently handled?” (their question 23), Nimbus/FTB replied “The contract has been entrusted to a company run by and employing accredited conservators.” This circular defence would suggest that the restorers, even while learning on the job, were judged capable of monitoring their own performance as well as the performance and the safety of their untested and still evolving methods.

These “fact sheets” contain contradictory material. They give an account (in answer to their question 16, “Has the formulation been changed?”) of the manner in which Arte Mundit’s composition had been changed after the 1999 approval of its “optimal” formulation: “In the first fifteen months of using the material the concentration of ammonia was less than 0.5%. Further development of the product allowed to [sic] reduce the level of ammonia, which is now less than 0.005%.” This account prompts two concerns. First, the reduction of ammonia to one hundredth of its original levels is not confirmed in official restoration documents (see below). Second, in answer to question 16 it was also stated that, regardless of the claimed dramatic reductions of the ammonia level, “The level of EDTA was not changed and the efficacy of the product remained identical.” This beggars belief: there are two cleaning agents in the paste, one an alkali, the other an acid – ammonia and EDTA. If the former was reduced to a hundredth of its original level, how could the efficacy of the whole not have been diminished on the one hand, and slewed in its pH composition, on the other? And, for that matter, what is the level of the EDTA? There is no confirmation in the “fact sheets” of Mr Stancliffe’s and Mr Odger’s joint claim in the December 2002 Conservation News that EDTA was used at 0.2%. In answer to question 4 (“What are its [Arte Mundit’s] constituents?”), EDTA is mentioned as a component but no figure is given for it. In answer to question 7 (“Are there different types?”), it is said: “Yes there are five different types, Arte Mundit I, II, III, IV and V. These are similar except the concentrations of EDTA differ with the lowest concentration (less than 2.5%) being Type II and the Highest concentration being Type V. Arte Mundit I contains no EDTA.” So what type was being used at St Paul’s? To find an answer we must turn to question 15 (“What type is used at St Paul’s?”) where it is revealed that: “After tests carried out at St Paul’s Arte Mundit V was formulated by Dr Eddy de Witte to address the specific conditions found at the Cathedral.” But on this answer we learn that the type of Arte Mundit which contains the highest levels of EDTA (at up to 10%), Type V, was the very one that had been specifically developed for St Paul’s – so from where does the figure 0.2% derive? Had EDTA been required only at the Stancliffe/Odgers claimed level of 0.2%, it would surely not have been necessary to develop a special type of Arte Mundit at all, because the already existing Type II, containing EDTA levels of up to 2.5%, would more than have sufficed?

When a cathedral worker whose station was next to a cleaning area complained to the Clerk of the Works on May 13th 2002 that strong smells were affecting her throat, he prepared a report (see Fig. 13) on May 17th saying that he had asked Nimbus to “get details of material, and improve ventilation”. On that day, she fell sick and was off work for a fortnight with a blocked nose and a bad chest having reportedly been told by the Clerk “Don’t worry, whatever they were using has been banned, they shouldn’t have been using it”.

In the following August a Health and Safety Executive COSSH ASSESSMENT (Control of substances harmful to health – see Fig. 14) identified the main active components of the Arte Mundit paste as: “EDTA (<30%); ammonia (<0.5%)”, which is to say with ammonia still at its original (and not the claimed massively reduced levels) but with EDTA levels then at up to an astounding 150 times higher than Stancliffe and Odgers were to claim publicly in December 2002.

The consequences of exposure to the EDTA-laced Type V paste, as stated on FTB Restoration’s own Material Safety Sheet of August 16th 2001, (where EDTA levels were already put at 10% and not at the later claimed figure of 0.2% – Figs. 11, 12 & 13) were said to be: “irritating to eyes, respiratory system and skin”. The primary route of exposure was: “Skin and eyes contact. Vapours inhalation.” The symptoms relating to use were: through inhalation – “Sore throat. Cough. Shortness of breath”; by skin contact – “Redness”; by eye contact – “Redness, pain. Tears”; by ingestion – “Abdominal pain, nausea.”

The staff member who had requested the safety sheet had recorded her own afflictions as they occurred. They made disturbing and progressively grimmer reading (– see Fig. 13). Because that member of staff had previously suffered from skin ailments, a specialist medical examiner who had been hired by the cathedral and who, (like the HSE inspector mentioned above) had accepted the claim that “the only change that was made [to the Arte Mundit] was to reduce the level of ammonia from 0.5% to 0.005% because of ‘the slight smell of ammonia that was present after the initial application’”, contended that her afflictions could not, “on the balance of probabilities”, safely be attributed to airborne chemicals in the cathedral. Nonetheless, he admitted that his decision might have to be reconsidered were “compelling further evidence in favour of occupational causation to be adduced”. This staff member had not been alone in her afflictions. In January 2003, a Press Association article (“Cathedral staff ‘have symptoms of chemical poisoning’”) reported that:

“Staff at St Paul’s Cathedral have been falling ill with symptoms of chemical poisoning, it emerged today. The Health and Safety Executive is sending investigators to the London landmark after staff reported suffering chest pains, respiratory problems and skin complaints. Chief suspect is the substance being used to clean the stonework of the historic building as part of a £40million restoration project. Arte Mundit, a cream paste that removes stains and dust on most surfaces, is being sprayed on to the fabric of the building. It contains ammonia, the smell and intensity of which has prompted the cathedral authorities to carry out all spraying at night. The man in charge of the project, surveyor to the fabric, Martin Stancliffe, was not available for comment today…Another substance in the cleaning mixture that might be causing the health problems is latex – it can cause skin allergies, sneezing, throat irritation and asthma…St Paul’s Cathedral registrar John Milne said that 20 out of 150 people working at the Cathedral had reported conditions which might be related to the restoration…”

In the March 2003 Conservation News, the St Paul’s/FTB Restoration method (as described there by Stancliffe and Odgers in December 2002) was challenged by conservation specialists. Professor Richard Wolbers, conservation scientist and solvents expert at the Winterthur Museum and Gardens, University of Delaware Art Conservation Department, charged the authors with appearing “not to understand very well the chemistry of the materials they are using” and of seeming to “resort to what I would call several common (and spurious) arguments to rationalise the ‘safety’ of their cleaning systems over other methods cited.” For example, he wrote, “EDTA is one of the strongest chelating materials one could bring to such a surface.” It would certainly dissolve “any calcium carbonate beneath it in the stone substrate that it may come into contact with…It is almost as if they simply adopted a commercial material that was easy to obtain or apply without considering what specific chemistry they were bringing with it to the stone surfaces or how it might affect the other constituents they might be adding.”

Professor Wolbers was highly critical of a number of other technical features of the programme but reiterated his fear that the authors “seem to have taken a poorly characterised material, a latex paste, and modified it with the addition of a considerable amount of EDTA (largely as an adaption in their minds, I suppose, of one of the main ingredients in the Mora’s AB57 cleaning system).”

John Larson, Head of Sculpture and Inorganic Conservation at the Conservation Centre, National Museums and Galleries on Merseyside, said that applications of moulding materials had contributed so much damage over the past 200 years that museums around the world “have now banned” their use, and that the application of liquid latex by brush or spray “has a dramatic effect on porous material such as stone…as it dries latex shrinks and clings tenaciously to the surface.” The effect of pulling it off the stone “exerts strong mechanical forces on the surfaces when the stone is carved and deeply undercut, as shown on the cover of Conservation News.” (See Figs. 1 & 2.)

Seemingly in the face of such attacks, accounts of the St Paul’s/FTB Restoration method shifted once more. In the May 2003 Conservation News Mr Stancliffe and Mr Odgers ignored Prof. Wolbers’ criticisms, which, they said, would be answered in a future article not by themselves but, instead, by the man who had developed Arte Mundit for the Belgian firm, Dr Eddy de Witte. Having said in December 2002 that “The original oil paint [of Sir Christopher Wren] is found to soften and can then be removed with water and scrubbing and this is both acceptable and desirable, as it is removing an unwanted and dirty paint layer”, Stancliffe and Odgers now insisted that Arte Mundit “is certainly not a paint stripper.” Apologising for having “misled” readers on the point, the pair claimed that when they had said “original oil paint” they had not been referring to the original oil paint but to “subsequent distemper applications and not to the original paint.” The distemper “is indeed softened by the latex”, they added, “as it would be by soaking with water”. At this point they admitted that the paste contained EDTA but gave no indication of whether it was at solutions of up to 0.2%, 10% or 30%.

A key concern of conservationists facing such methodological discrepancies was whether or not the EDTA migrated into the stone during the periods of curing after being sprayed as a water-bound paste on to porous surfaces that had already been attacked with caustics and abrasives by previous restorers. With regard to Prof. Wolbers’ fear that Arte Mundit’s EDTA would have the time and the opportunity to invade and damage the stone yet another set of Stancliffe/Odgers claims was revised. In their May 2003 account, they claimed that the latex solution was sprayed to a depth of only 2mm and left for only “two or three hours” when in December 2002 they had said that the curing lasted “usually 24-48 hours”. In May 2002, Conservation News reported that the latex was removed after “one to four days”. In April 2003, the BBC reported that the latex was left on the stone surfaces for between “One to four days [depending] on the thickness and temperature”.

In December 2002, Stancliffe and Odgers had claimed that after removal of the Arte Mundit paste and subsequent washing and scrubbing, the stonework “still retains a patina”. In May 2003 they admitted that John Larson had rightly pulled them up for their “inappropriate use of the word”. Today the surface of the stone can be seen to have been left porous, susceptible to invasion by pollution, and chalky. Its weakened surface can now be rubbed away with a wipe of dry cloth – see Figs. 15 and 16.

Michael Daley

Printable PDF version of this article:
ArtWatchUK_5_06_11_st paul’s 2.pdf

Comments may be left at: artwatch.uk@gmail.com