New Contributions to Iron Gall Ink Inspection Protocols Using Open Source Surface Analysis and Digital Imaging

1.1 The Deterioration of Iron Gall Inks

Documents written in iron gall ink are cultural heritage objects that are susceptible to deterioration. This is due to the composition of the ink. Iron gall inks are formulated from three main components: tannins from oak galls, ferrous sulphate and gum arabic as a binder, in a liquid medium which may be water or wine (Díaz Hidalgo et al. 2018; Eusman 1998; Karnes 1998; Kolar and Strlic 2006; Stijnman 2002). In some cases other components such as vinegar, pigments or pomegranate peel are added to modify specific properties such as gloss or colour or to improve preservation (Karnes 1998). The colour of the ink is due to the chemical reaction between the metal salt and gallic acid. The iron (II) ions oxidise in contact with oxygen and turn into iron (III) ions (Hahn et al. 2004; Neevel 1995); they become an insoluble pigment, while the binder keeps these particles in suspension (Kanngieber et al. 2009; Rouchon et al. 2011).

The problem with these papers is the corrosion of the inks which is caused by two degradation mechanisms (Duh et al. 2017; Gal et al. 2014; Viegas et al. 2013). The first is acid hydrolysis catalysed by the acidity of the ink. The second is the oxidation of cellulose catalysed by free iron (II) ions. Both mechanisms result in the breakdown of the cellulose polymer chain (Kolar and Strlic 2004, 2006; Reiβland 1999). In addition to composition internal factors, such as the amount of ink applied and the composition of the paper substrate, and external factors, such as changes in relative humidity, environmental pollution and document handling, contribute to this degradation process (Reiβland 2000).

Corrosion is evident in ink transfer, with colours ranging from light brown to more intense shades. The presence of halos around the inked areas is first seen under ultraviolet light with a greenish fluorescence. As the corrosion progresses they become visible with colours ranging from light brown to very dark brown. Finally, there is mechanical damage to the inked areas such as micro-cracks, fissures, cracks and loss of support. However, this corrosion process begins long before it can be detected by visual inspection (Reiβland 2000; Rouchon et al. 2009).

1.2 Inspection Protocols

Many institutions and researchers have been studying the problem of corrosion of iron gall inks for years. The Iron Gall Ink project, initiated in 1990 by a Dutch research group, focuses on the impact of iron gall inks on museum, library and archive collections. It examines the mechanisms of deterioration, historical recipes and treatment of documents with these inks. In addition, the project provides protocols for assessing the state of conservation, together with a risk assessment and possible treatments (The Iron Gall Ink website 1998).

At the same time, the Instituut Collectie Nederland (ICN) investigates the causes and mechanisms of degradation of these inks in order to develop conservation treatments. They rate the visible progress of ink degradation on a scale from 1 to 7 and take into account factors such as fluorescence, verso markings and microcracking. Interpretation on original documents is complex due to possible variations. Therefore, the visual examination is carried out on the most affected area of the verso of the document. The CIE proposes a classification of four document conditions. Condition 1 has no discolouration or is very light brown in colour, where the condition is good and the document can be manipulated. Condition 2 has dark brown discolouration without mechanical damage, where the condition is fair and the document should be handled with care. Condition 3 has mechanical damage such as cracks, where the condition is poor and requires expert handling. Condition 4 has severe loss of support, where the condition is poor and requires careful handling by museum or archive staff to prevent further loss (Reiβland and Hofenk de Graaff 2001).

Subsequently, in 2008, Sylvia Albro and researchers at the Library of Congress, Washington DC, advanced the examination of these documents and introduced Protocols for Iron-Gall Ink Treatment (PIT) (Albro et al. 2008). They use standardised documents to record the results of visual and chemical examination together with a glossary of degradation and paper support characteristics. They also include two treatment charts with a comprehensive approach to determining the most appropriate treatment for each document (Albro et al. 2008; Dekle and Haude 2008).

The problem of corrosion of iron gall inks also affects artistic drawings. Joanna Kosek and Caroline Barry of the British Museum use in-house surveys and adapt protocols from the ICN, the Library of Congress in Washington DC and the Iron Gall Ink website. The aim is to create a quick and easy system for assessing the condition of artistic drawings using iron gall ink. They use a Microsoft Excel form as a database to record each deterioration and classify the ink condition of each drawing analysed. This facilitates the recording of relevant information for each object. In addition, colours are used to indicate the four ICN conditions (Kosek and Barry 2019).

In summary, inspection methods rely on a variety of tools such as identification, direct observation, imaging techniques and chemical tests to diagnose deterioration. This research has been a major breakthrough in the field of documentary heritage. It has contributed significantly to the conservation of historical documents with iron-gall ink in various institutions. Our research aims to contribute to these inspection protocols by adding digital tools to assess the progress of ink corrosion and analysing support loss and halos in inked areas.

1.3 Digital Image Analysis in Heritage Research

ImageJ^® software (developed by Fiji) is an open-source digital image processing and analysis program. It is designed for the study of scientific images particularly in the fields of biology, medicine, engineering and others. Within cultural heritage it has been applied to the study of rock art but there are not many studies in other disciplines (Ferreira and Rasband 2012; González 2018; Schneider, Rasband and Eliceiri 2012).

This programme has an intuitive interface with well-organised tools and menus and a very comprehensive manual. In addition, it supports several image formats, from the most common such as TIFF, JPEG and PNG to microscopy specific formats such as LSM and ND2. ImageJ^® allows brightness and contrast adjustment, rotation, cropping, filtering and histogram analysis, segmentation and feature measurement. It also has tools for particle counting, measuring areas and perimeters, length and intensity in an image. This enables quantitative image analysis (Ferreira and Rasband 2012; González 2018). An important advantage is the ability to batch process images. This speeds up the analysis of large datasets and reduces turnaround time.

Plugins can be added for specific and specialised tasks. For example, DStretch^®, developed by Jon Harman in 2005, is widely used for rock art images, especially pictogram remains (Harman 2008). It is a filter correlation tool used to recover nearly lost rock paintings and engravings. By correlating the colour channels of the image, certain pigments can be enhanced. This allows almost lost pictograms to be read (Le Quellec et al. 2015). Its use in conjunction with technological advances in gigapixel imaging offers optimal results (Quesada and Harman 2019). In mural painting, for example, it is possible to virtually reconstruct missing paintings from the interior of a church (Fuentes, Soto-Martín and Martin-Gutierrez 2020). It can also be used to transpose decorations on ancestral textiles to recover motifs of heritage interest (Lungu et al. 2021). In metallic archaeological heritage it is used to obtain a three-dimensional graphic of the surface of bronze pieces from a multifocal optical microscopy image. The aim is to obtain information about ancient manufacturing techniques and the presence and distribution of inclusions in the metal (Figueiredo et al. 2013).

However, there is little research on the application of ImageJ to documentary heritage. One study shows the use of the software to examine the grains of the drying agents used in inks and to estimate roundness, circularity and morphological parameters. The aim is to characterise the drying sands used in inks during the Portuguese Inquisition (Nunes et al. 2023).

3.1 Case Study. Documents From the Sixteenth Century in the Archivo Histórico Provincial de Santa Cruz de Tenerife

The Archivo Historico Provincial de Santa Cruz de Tenerife (AHPT) contains a large number of documents with iron gall ink. They belong to the public sector, such as documents from confiscated convents, notarial records and judicial documents, and to the private sector, such as personal and family archives. Due to the large number of documents, the inspection protocol is applied to a selection of documents from the 16th century, the oldest documents with these inks in this archive.

The following funds are examined: Mortgage Accounts; Seized Convents; Ancient Regime: Judiciary; and Historical Section of Notarial Protocols. The study protocol proposed by the ICN is applied. Each folio is classified according to one of the four conditions. In addition, notes are made of the damage found and the visual appearance of the iron gall ink. Photographs are taken of the documents both in general and in detail. Twenty folios are selected from each fund (Table 1).

3.2 Documentation and Data Collection

A Nikon D850 DSLR camera with a 50 mm fixed lens is used in the zenith position with diffuse lighting from two light sources positioned at 45° to the model. A ruler is used as a measuring reference and the Kodak grey scale. Photographs are also taken using a calibrated DinoLite AM3113T digital optical microscope, with magnifications of 50× and 200×.

3.3 Processing and Study of Digital Images

The tools and their use in the study of documents with iron gall inks used in this research are shown in Table 2.

To calibrate a digital image in ImageJ, proceed as follows:

1. Draw a line of known length on the image ruler and select the Set Scale tool (Analyse menu).

2. A pop-up window opens. In Known distance enter the measurement made on the ruler; in Unit of length change pixels to millimetres. Accept and you have the calibrated image.

To obtain actual measurements of support loss or the area affected by corrosion the image must be calibrated and a ruler used as a reference. The Scale Bar tool is used (Analysis menu, Tools). The colour, thickness, font size and position of the bar are selected in the pop-up window. A density bar can also be added to the image using the Calibrate Bar tool. This is a scale of the grey tones present in the image.

The Colour Threshold tool allows you to threshold or segment the image to select specific areas of interest such as document loss. Together with the ROI Manager tool you get several useful options. The main one is the ability to register regions of interest and extract measurement data from the selections (areas, perimeters). You can also number and label each of the losses, automatically or manually, and obtain measurements from each of the selections. It is also possible to generate a vector image of the image selections, such as a mask image (Figure 1). This tool is an interesting resource for calculating the percentage of document loss. The vector image can also be used for damage maps.

Figure 1:

Using the ROI Manager tool to record support losses due to corrosion; a table with information on the areas, perimeters and X and Y coordinates of the selections.

The Plot Profile tool generates a two-dimensional graph of pixel intensity along a line drawn in the image. The Surface plot tool generates a three-dimensional graph of pixel intensity within a rectangular selection in the image (Figure 2). The X-axis represents the distance along the selection. The Y-axis represents the pixel intensity. To use this tool images must be greyscale (8, 16 or 32 bit).

Figure 2:

The tools are tested on a grey scale to visualise the results. The top graph is created using the Surface plot tool. It plots the grey levels of a rectangular selection on three axes. The lower graph is created using the Plot Profile tool. It plots the grey levels in the peaks of a linear selection of the image on two axes.

These tools make it possible to examine the surface of iron gall inks with halos in the inked areas. The presence of halos is a sign of degradation of ferro-gallic inks. It is caused by the oxidation of the cellulose in the paper. They are caused by the migration of soluble acid components – sulphuric acid present in the ink – into the surrounding areas, affecting the substrate. This migration is accelerated by high humidity, a previous incident – leakage, flooding – or inadequate water treatment (Reiβland and Hofenk de Graaff 2001). In turn, the iron ions in the ink are oxidised and transformed into ferrous ions which are also visible as dark halos in the surrounding areas (Neevel and Mensch 1999).

The calibrated image provides real measurements of both the ink stroke and the halos. So documents with halos in the inked areas are selected, with the dimensions checked with both tools. The intention is to find a relationship between expansion and the level of corrosion. The graph obtained represents the intensities of grey along which the line is drawn. This provides information on the shades of the halos, the ink line and the support, identifying the shade and distance of each area (Figure 3).

Figure 3:

Application of the plot profile tool to a detail of folio 924 of document PN201. The graph shows the highest peaks representing the paper support, the lowest peaks representing the ink stroke and the average peaks representing the halos.

Another possibility of this tool is to study the homogeneity of an inked area. A three-dimensional graph of a selected surface is obtained (Barman et al. 2021; Ferreira and Rasband 2012). Figure 4 shows an example of the type of graph and the reading obtained. We can distinguish the peaks corresponding to the paper support and the peaks corresponding to the ink stroke.

Figure 4:

Applying the Surface plot tool to two details of iron gall ink strokes on the front page of document CH1.

In addition to all the tools and possibilities offered by the ImageJ software the Dstretch plugin is explored. This plugin, by correlating filters, highlights traces of pigments that help to read almost lost works of art. It is tested on documentary heritage to evaluate its possible applications. The image is opened in the software and the DStretch plugin is selected. A pop-up window opens allowing a series of filters to be tested and the intensity to be adjusted. Each of the filters gives a different result (Figure 5).

Figure 5:

Application of the DStretch plug-in to a detail from folio 2 of folder 8 of Judicial Document 1976A. The document shows fractures and loss of support; it is in poor condition or 4 according to the ICN classification. (a) DStretch plugin pop-up window. Bottom, menu with default decorrelation filters. (b) Results obtained from the document with several of the default filters. From left to right, top: CRGB, YUV and LYE filters; below: YRE, LABI and RGB0 filters.

4.1 Documentation and Data Collection

By applying the Colour Threshold and ROI Manager tools to the 20 selected folios from each fund accurate data is obtained on the total loss of support due to corrosion of the iron gall inks. Masks were also generated to facilitate the production of damage maps (Figure 6).

Figure 6:

The image shows three examples of documents with support losses. On the left is the image of the document with the loss selections in red; on the right are the generated masks.

The results show that the documents of the ancient regime: Judicial Fund have multiple small losses affecting the entire surface. Therefore, they have a larger affected area. The documents in the Historical Section of Notarial Protocols have large losses and other smaller ones. The documents in the Mortgage Accounts fund have fewer losses but are large in size. This indicates a state of severe corrosion and fragility that makes them difficult to handle. For this reason access to them is restricted. Finally, the documents in the Seized Convents Fund have small losses in certain areas.

The results show the extent and distribution of paper losses in each fund. This enables action to be planned (Table 3).

4.2 Halos in Inked Areas

The Plot Profile and Surface Profile tools are used to measure the diffusion of the halos (Figure 7). Lines are drawn in different areas of the ink strokes with halos to create a range of measurements. The results obtained in the examples in Figure 7 are as follows: in the first image, two measurements are shown, which shows that the range is between 1.3 and 1.4 mm. In the second image the range of the halos is between 1.2 and 1.8 mm. These measurements show the maximum and minimum values for each document fund (Table 4).

Figure 7:

Application of the plot profile and surface profile tools and the resulting graphs. The documents are: (a) extract from page 864 of document PN201; (b) extract from page 812 of document PN201.

The results show that the documents from the Ancient Regime Fund: Judicial and the Historical Section of the Notarial Protocols have more extended halos, up to 2 mm. On the other hand, the documents from the Seized Convents have smaller halos, which is predictable given their better state of preservation.

However, as can be seen in the table, all the funds analysed have documents with visible halos in some lines. This indicates a process of degradation of the iron gall inks. Regular inspections are essential to identify the problem and determine which documents require priority action.

The appearance of halos – initially as a greenish fluorescence, followed by a light to dark brown colouration – indicates a level between 2 and 5 on the ICN scale (Reiβland and Hofenk de Graaf 2001) but there is no correlation between the extent of the halos and the level of degradation of the document.

4.3 Reading Faded Texts With DStretch

The DStretch plugin offers several preset filters. In order to verify its usefulness for document preservation, tests are carried out. Documents with deteriorations such as dark bleeding, halos in inked areas, ink fading, etc. are selected. Documents that are difficult to read due to deterioration.

The LYE filter is particularly effective at converting yellow tones to dark brown. It gives optimal results on faded text by intensifying the ink (Figure 8b). On the other hand, for documents with oxidised and darkened inks, the YBG and LBK filters are effective. They intensify the ink strokes and lighten the paper support (Figure 8d and e).

Figure 8:

Application of the DStretch plugin to two documents. (a) Visible image of folio 1 of document CH-1; (b) application of the LYE filter; (c) visible image of folio 2, folder 7 of judicial document 1976A; (d) application of the YBG filter; (e) application of the LBK filter.

In addition, the LYE filter gave an interesting response in one of the documents. It has ferro-gallic inks in two shades, a light brown and a black, probably due to the composition itself. This filter detects them with a white or pinkish colour (Figure 9b). In this case it would be advisable to analyse the composition of the inks to study the response of the ink with this correlation of filters.

Figure 9:

Detail of Folio 5 of document Convento-1072. (a) Visible image; (b) application of the LYE filter.

DStretch was useful for documents with advanced corrosion that made reading difficult and for documents with text loss due to ink fading. The LBK and YBG colour spaces gave positive results, reducing the intensity of halos and bleed-through and enhancing text. The LYE colour space is a very versatile filter. On the one hand it changes yellow tones to dark brown which works well with faded text while, on the other hand, it changes the blackest inks to white or pink tones.