A programme opens at Chartawell when a question keeps arriving from collection holders and cannot be answered from the literature. It closes when the data is released, whether or not the answer is interesting. Three of the six programmes below have so far produced null results.
Inks and media
Iron gall ink is the reason a great many manuscript leaves are now unsafe to turn. Free acid and iron(II) attack the cellulose directly beneath the stroke; the leaf browns, then perforates along the writing itself. The mechanism has been understood since the 1990s. What is still argued about is prediction: which leaves will fail in the next decade, and whether any of the non-destructive indicators proposed for triage actually rank leaves correctly.
Our current survey has measured 1,140 leaves from estate and legal papers across eleven repositories, scoring each for stroke-side discolouration, transmitted-light perforation and bathophenanthroline response, then re-measuring the same leaves at eighteen-month intervals.
| Indicator | Correlation with observed loss | Practical verdict |
|---|---|---|
| Bathophenanthroline indicator paper | Strong for free iron(II), weak for rate of loss | Use to confirm presence, not to rank urgency |
| Transmitted-light perforation count | Strong | The most useful single field measure |
| Visual browning score | Moderate, and heavily reader-dependent | Only with two independent scorers |
| Cold extraction pH of adjacent blank area | Weak | Does not represent conditions under the stroke |
The uncomfortable part of that table is the last row. Whole surveys have been designed around measuring pH somewhere convenient on the leaf, and it does not tell you what is happening where the damage is.
Collection environments
Most guidance on store conditions was written when the plant ran continuously. It now often does not. The question we were asked repeatedly — and could not answer from the literature — is what a nightly setback of three or four degrees does to bound volumes over a period of years, as distinct from what it does to a test strip over six weeks.
The programme runs as a monitoring loan. Partner archives take a logger set for a full annual cycle, with paired sensors inside a bound volume and in free air on the same shelf. We analyse; the archive keeps its own data and can publish it independently of us.
- Nineteen stores instrumented since 2023, of which fourteen have completed a full year.
- Interior-of-volume RH lags free air by between four and thirty-one hours depending on binding.
- In eleven of fourteen stores, the daily cycle at the text block was under two percentage points even where free air moved by eight. The mass of the collection is doing most of the buffering.
- The three exceptions were all rooms with a fan coil discharging directly onto shelving.
Fibre furnish and permanence
Between 1860 and 1935 British mills moved from rag through to bleached chemical wood pulp, with a long, messy overlap in which both appear in the same ream. Cataloguers routinely describe an item as “wood pulp paper” on the evidence of colour alone, and are wrong often enough to matter when the description is later used to prioritise digitisation.
We are building a reference set of identified furnishes with dated provenance: 2,300 samples so far, each with Herzberg and Graff “C” stain photomicrographs, lignin spot test, cold extraction pH and, where a duplicate exists, fold endurance after accelerated ageing.
The other three programmes
Smaller, and reported here rather than given a page of their own until there is something to say.
| Programme | Opened | Status |
|---|---|---|
| Mass deacidification: efficacy after ten years on the shelf | 2021 | Null result. No measurable difference in fold endurance between treated and untreated sheets from the same batch after a decade, in a sample of 140 leaves. Written up in full. |
| Handling damage under digitisation | 2024 | Open. 42 bound volumes photographed before and after a scanning campaign, scored for spine and hinge change. |
| Adhesives used in twentieth-century repairs | 2025 | Open. Identifying what was actually used, since repair records almost never say. |
With the three above, that is six open programmes, of which three — the nightly setback study, the mass deacidification review and the 2022 ageing round robin — have so far produced null or negative findings.
Released datasets
Closed programmes are released with raw measurement files, instrument settings and the analysis scripts, not only the summary tables. If our conclusion is wrong, that should be demonstrable from the same files we used.
| Dataset | Released | Contents |
|---|---|---|
| Machine-made furnish reference set (partial) | 2025 | 2,300 samples, photomicrographs, stain results, pH |
| Mould recovery after wetting incidents | 2024 | Two incidents, 26 months of post-event monitoring |
| Surface cleaning trials | 2023 | Six materials, gloss and colour change, abrasion counts |
| Accelerated ageing round robin | 2022 | Four laboratories, identical stock, divergent results |
Collaboration and visiting places
We hold two visiting places a year, each of three months, for conservators and scientists who want bench time on a question of their own. There is no fee and no obligation to publish with us. Applications are read twice a year by the research office; a proposal that mainly needs an instrument we happen to own is a perfectly good proposal.
Joint funding bids are welcome where the centre’s role is measurement and reporting. We do not take work from suppliers of conservation materials or environmental plant, and we say so in writing at the start of any bid, because the value of a result from here depends entirely on our having no stake in which way it comes out.