1 Corpus

The seventy-eight manuscripts that currently comprise the Jazzār corpus are scattered around the world but are identifiable by three means of library annotation: endowment statements, endowment seals and short endowment mottos (on these elements, see the Introduction to this volume). The first chronologically datable activity in the Jazzār corpus was the annotation of the endowment statement at the beginning of the volumes. The statements are dated between 1196/1782 and 1205/1791, with a striking majority written in the years 1196/1782–1197/1783 by the same hand. While the layout and number of lines can vary, the wording is standardised, although a few variants have been noticed. The second datable marking method consists of seal impressions. They can be found at the beginning of the volumes, either on the first folios or on the recto (corresponding to side a) prior to the beginning of the text. The black-inked impressions were left by one of two different matrices, one circular and one oval, both dated 1205/1790–1791, as can be seen in the lower margin of the impression.⁴ It is possible, however, that the seals were used beyond this date.⁵ The third method, the annotation of mottos, is not dated. This extremely short version of the endowment statement can be found on the margins of several pages throughout the book. If there are only two textual variants for the mottos—A) ‮وقف الله تعالى‬‎, which is the most frequent, and B) ‮وقف لله تعالى‬‎—everything else differs greatly: the location in the page and their frequency throughout the book, the writing implements and inks used, and, even more surprisingly, the writing hands and styles. Multiple series of mottos by different hands were observed in the same volume (Ar. 3272, for instance), but they can also be completely omitted (as in Ar. 3316). Said Aljoumani’s preliminary palaeographic assessment identifies around twenty-five distinct hands and subtypes of hands that penned mottos and endowment statements in the manuscripts identified so far.⁶ Additionally, in this case, one hand—from now on referred to as Hand no. 1—is found more frequently and corresponds to the same one that wrote most of the endowment statements from 1196/1782 to 1197/1783, suggesting that it might belong to the first librarian. Finally, single occurrences of mottos were also found: they are often incomplete, shorter versions of the usual formula and are sometimes written by unskilled hands. For this reason, they might be errors, pen trials or writing exercises.

A subset of nine al-Jazzār manuscripts are preserved at the CBL, all identifiable from the library inventory written in 1221/1806.⁷ As has been mentioned previously, seven manuscripts were analysed for this paper because the other two were unknown to us at the time. The most important characteristics of the manuscripts, particularly referring to the three marking methods, are summarised in table 11.1.

Two manuscripts of the CBL corpus—Ar. 3342 and Ar. 3272—present the standard endowment statement, written by the most frequent hand (Hand no. 1) and dated 1196/1782 and 1197/1783, respectively. During the year 1197/1783, a different hand (Hand no. 2) penned a shorter version of the endowment statement on Ar. 3334. A third, different hand (Hand no. 3) wrote the standard version of the endowment statement on Ar. 3315; the annotation is undated, but it overlaps the seal impressions, suggesting it was written after the seal was stamped. Five of the manuscripts do not present any endowment statements.

The seal impression is visible on eight of the CBL manuscripts, although in one case (Ar. 3310), it was partially erased in an attempt to mask the provenance of the book. The circular matrix was mostly used to mark these manuscripts, with the oval seal appearing in only two of them (Ar. 3236 and Ar. 3342).

Several hands penned the mottos in the CBL manuscripts. Hand no. 1 appears in two manuscripts, showing slightly different writing styles on the upper margins of Ar. 3268 and on the upper and fore margins of Ar. 3272. In both cases the variant A of the motto is repeated on several folios throughout the books, at almost regular intervals (on average, every 20 and 30 folios, respectively). Both books, however, also show mottos written by other hands. In addition to two incomplete mottos (U2 and U3), two sets of mottos were left by Hand no. 4 (motto B) and Hand no. 5 (motto A) in Ar. 3272. They occur at more irregular intervals than Hand no. 1, but with a certain frequency. On the contrary, Hands no. 10 and 12 only added two (motto A) and one annotation, respectively, to Ar. 3268. Hand no. 10, which shares some palaeographical characteristics with Hand no. 1, wrote variant A mottos on the upper margins of fols 1^b–2^a in Ar. 3342. The person (Hand no. 3) who wrote the endowment statement on fol. 1^a of Ar. 3315 also added a motto (variant A) on the same folio, between the decorative cartouche and the statement. Another hand (no. 8) added mottos on the upper margins of several pages of the same volume, at irregular intervals. Hand no. 7 penned mottos (variant A) on the upper margins on the verso (side b) of several folios of Ar. 3236, also at irregular intervals, while the motto (variant B) left by Hand no. 6 and an incomplete one (U1) are visible on the fore margin of fol. 1^a. Ar. 3334 sports only a series of mottos on the upper margin made by Hand no. 9, while Hand no. 11 wrote two mottos (variant A) on the upper margin of fol. 30^b and lower margin of fol. 31^a in Ar. 3310.

We provide a summary of the writing hands encountered in the CBL corpus with examples of their writings in table 11.2.

2 Ink Classes and Scientific Analysis

Black inks can be grouped into four main classes: carbon, plant, iron-gall and mixed inks.⁸ Carbon inks are obtained by mixing charcoal or soot with a water-soluble binder in a water-based solvent. Several precursors were used to obtain the carbonaceous material in the Islamic world: lamp oils were the most appreciated, due to the fine particles of the soot obtained, although the burnt material collected from ovens and pots was also used because of its abundance and convenience, despite being of a lower quality. Gum arabic is normally used as a binder dissolved in simple fresh water or enriched with some perfumes, such as rosewater. Carbon inks can sometimes contain metallic impurities, such as traces of iron, copper or lead, probably contaminants from the tools used to prepare or store the ink (such as bronze or lead inkwells) or from the water. Such elements are present in a more consistent amount in rare cases, suggesting an intentional addition rather than an occasional contamination. We refer to the latter as carbon-based inks with metallic admixtures, and they are sometimes included in the class of mixed inks.⁹

It is possible to obtain tannins, which are the main components of plant inks, by cooking or macerating various vegetal materials, such as bark, fruits or flowers. There are only a few Arabic recipes for plant inks, and they were rarely employed, probably because the final product is not black but rather light to dark brown. Tannins, particularly hydrolysable ones, are also among the main components of iron-gall inks. The latter are obtained by the reaction between iron ions (Fe²⁺) and gallic acid in a water-based solvent, with the addition of a binder (gum arabic).¹⁰ Arabic recipes mention gall nuts as the best source for gallic acid, but a variety of plant parts (for example, tree bark, leaves and fruit) can be used as substitutes. The most common and appreciated source of iron is vitriol (a mixture of hydrated metallic sulphates, often including copper and zinc). Filings, slag, nails and pieces of iron were used in a small number of Arabic recipes, resulting in inks without metallic impurities, although the extraction of iron ions and the consequent reaction are more difficult to obtain.¹¹

The class of mixed inks comprises the inks resulting from blending together inks or ingredients from the previous classes. Several mentions of these preparations can be found in Arabic treatises. This class consists of two subtypes: mixed carbon-plant inks, when tannins are added to carbon inks, and mixed carbon-iron-gall inks. The latter include inks with different ratios of carbon to iron because they can be formulated by either adding small amounts of carbonaceous material to an iron-gall ink or including some vitriol and tannins in a carbon ink. Such inks are attested for the first time in Arabic recipes dating between the ninth and thirteenth centuries and have continued to be used until recent times.

In order to differentiate the inks, we exploit the distinct optical properties that their ingredients have in IR light: carbonaceous materials remain opaque throughout the IR regions, unbound tannins appear transparent at around 940 nm, while the iron-gallotannic complex, characteristic of iron-gall inks, loses opacity more slowly and become completely transparent between 1200 and 1500 nm.¹² We can observe the presence of tannins in UV light, as they have the ability to quench the fluorescence from the writing material, thus, enhancing the contrast between the ink strokes and the background. This behaviour is readily apparent in pure plant and iron-gall inks. In mixed ink, however, it can be difficult to notice because the carbonaceous materials are opaque in UV light and can, therefore, mask the enhancement of the tannins unless the amount of carbon particles is negligible, the writing is damaged or the tannins spread into the writing support around the stroke.

The XRF spectrometry is used to identify the elemental composition of the inks. Semi-quantitative measurements can also be used to differentiate inks, especially of the iron-gall type, based on the various ratios of the vitriolic impurities (copper, zinc and the like) to iron (the main component).¹³ This method was successfully applied to discriminate mediaeval European iron-gall inks,¹⁴ but it can also be effectively employed to differentiate mixed carbon-iron-gall inks when vitriol was used. It is also possible with XRF to carry out a qualitative evaluation of the impurities or the addition of metallic admixture in carbon-based inks.¹⁵

3 Analytical Protocol and Equipment

The inks have been studied according to the standard protocol for the characterisation of writing materials in use at the CSMC at Universität Hamburg.¹⁶ This protocol consists of a visual examination of the writing substances with a microscope in visible light, followed by a primary screening performed with NIR and UV reflectography at two specific wavelengths. IR reflectography is then used to capture images in the short wave IR region (900–1700 nm), while the elemental composition of the inks is determined with XRF. Additional in-depth analysis conducted with various spectroscopic methods can be added to the protocol but were not applied in this case.

The visual examination of the inks and the primary screening is conducted with a miniature, handheld Dino-Lite USB microscope. The microscope has built-in LED illumination at 395 nm (UV) and 940 nm (NIR), and a customised external white light (VIS) source functioning as a stand. In VIS light we observe the feature of the inks—that is, the distribution of the writing matter and the presence of particles, halos, cracks, bleeding or of other alterations of the writing due to damage, loss or retouching. By comparing the change of opacity of the writing between the visible and the IR micrograph, it is possible to differentiate the pure classes of black inks. Additionally, the presence of tannins can be detected in many cases by examining the micrograph captured under UV light: tannins have the ability to quench fluorescence, thus enhancing the contrast between inks with tannins and a fluorescent background under UV light. In the case of tannins spreading in the writing support beyond the limits of the ink stroke, it is even possible to observe that its dimensions appear larger in the UV micrograph compared to the VIS image.

OPUS Instruments APOLLO Infrared Reflectography Imaging System is used to ascertain the presence of carbon in the inks and allow for the discrimination of pure iron-gall inks and mixed inks. The regular short-wave IR sensing range (900–1700 nm) of the 128 × 128 pixel scanning InGaAs sensor can be reduced by the following filters: short wave pass filter (SWP1250, range 900–1250 nm), band pass filter (BPF1250–1510, range 1250–1510 nm) and long wave pass filter (LWP1510, range 1510–1700 nm). Each filter is mounted in front of the IR lens (150 mm, f/5.6–45). The working distance between the sensor and the object for this analysis was set to 74 cm with an aperture of f/11 and exposure time of 50 ms per tile. Two 20 W halogen lamps provide broadband illumination. In order to carry out this study, we used the long wave pass filter to limit the range of IR light to the portion of the spectrum where iron-gall inks become completely transparent.

Elio Bruker Nano GmbH (formerly XG Lab) is used for the identification of the elemental composition of the inks and was obtained through this compact X-ray spectrometer. It features a 4 W low-power rhodium tube and adjustable excitation parameters. It has a 17 mm² silicon drift detector (SDD) with energy resolution < 140 eV for Mn Κα. The beam size is roughly 1 mm, compatible with the size of the ink strokes we selected for the analysis. The measurements were performed on single spots at 40 kV and 80 μA, with an acquisition time of 2 min. At least three spectra for each different writing were acquired. Bruker’s SPEKTRA software was used for the peak fitting and the semi-quantitative data evaluation.

4 Results and Discussion

The results of the analysis on endowment statements and mottos are summarised in table 11.3.

As shown in this table, different classes of inks were identified, although the majority of them belong to those of pure carbon inks and mixed carbon-iron-gall inks with a prevalence of the carbonaceous component. We observed a pure iron-gall ink only in the endowment statement and motto written by Hand no. 3 on the first folio of Ar. 3315. We tentatively assigned the inks to the classes of mixed carbon-plant ink in two cases and to the subclass of carbon-based ink with metallic admixture in one case. The dubious identifications are marked with a question mark in table 11.3. Such uncertainty arises from the impossibility of detecting the organic component of the ink with the methods applied.¹⁷ Additionally, the manuscripts were subjected to heavy restorations in the past. The application of aqueous treatments, splitting and lamination, and pigment retouching have caused the migration of ions (including iron ions) from the original inks to the paper and the introduction of external materials and elements. This made the evaluation of XRF data particularly complicated, especially when dealing with elements detected as traces in carbon-based and mixed inks, because the high amount and heterogeneity of the same elements in the writing support makes it hard to discern whether they are impurities in the inks or a contribution of the writing support.

Despite these difficulties in analysis, we can, nevertheless, observe that inks belonging to the same class are also very different due to their optical properties, elemental composition and, for mixed inks, the ratio of carbon to iron-gall. The latter can only be qualitatively appreciated in the VIS and IR images. The distribution of mixed carbon-iron-gall and pure iron-gall inks depending on their elemental composition is shown in fig. 11.1. The graph displays the percentage of iron, copper and zinc detected in the strokes normalised to their total.¹⁸ It is possible to observe three subgroups, which suggests the use of different vitriols for the ink preparation. The first subgroup is characterised by a high amount of iron (65–90 %) and a low amount of copper (10–35 %) with no or just traces of zinc (0–12 %), corresponding to the inks used by Hand no. 2 to write the endowment statement on Ar. 3334 and by Hand no. 3 for both the endowment statements and motto on Ar. 3315 (highlighted by the light blue circle in fig. 11.1). The second subgroup is characterised by a high amount of iron (60–75 %), a low amount of copper (25–40 %) and zinc (13–25 %), corresponding to the inks used by Hand no. 9 to write the mottos on Ar. 3334 (highlighted by the light green hexagon with a dashed line in fig. 11.1). Finally, the third subgroup is characterised by a medium amount of iron (35–60 %) and copper (25–45 %) and a low amount of zinc (0–12 %), corresponding to the inks used by Hand no. 4 and Hand no. 10 to write the mottos on Ar. 3272 and Ar. 3342, respectively (highlighted by the yellow hexagon with a solid line in fig. 11.1). The ratio obtained for the spots analysed on the partial motto named U2, written on Ar. 3272 by an unknown hand, falls exactly in the same area occupied by results of the Hand no. 10, suggesting that it was written with the same ink. A closer palaeographical examination confirms that it was probably penned by Hand no. 10 and left incomplete perhaps for aesthetic reasons or for an erroneous positioning in the book (see table 11.2 for comparison). The ink used by Hand no. 6 and that of the motto U1 from an unknown hand, both penned on the fore-edge of Ar. 3236, show a broad fluctuation in the elemental content and cannot be assigned to any of the subgroups. As has been clarified previously, such inhomogeneity is probably caused by the spreading of metallic elements in the writing support during conservation treatment.

The only pure iron-gall ink was used by Hand no. 3 for both the endowment statement and the motto on fol. 1^a of Ar. 3315. Interestingly, this is the only hand that can be surely dated to after 1205/1790–1791, as it overlaps the seal impression. The identification of its class is based on optical properties in both IR and UV light (fig. 11.2a–b): the ink clearly changes opacity at 940 nm and becomes transparent above 1510 nm, and the presence of tannins makes it homogeneously dark in UV. The increased intensity of the peaks of iron, copper, zinc and sulphur in the XRF spectra of the ink with respect to those of paper confirms the identification as a vitriolic iron-gall ink (fig. 11.2c). Probably due to the damages that this ink suffered, the writing was retouched with a carbon-based ink in some places. This intervention is clearly visible in the images taken in IR light, as the overlapping ink does not change opacity (figs. 11.2a and 11.2d). As seen on fig. 11.1, the elemental composition of the ink used by Hand no. 3 is similar to the one used by Hand no. 2 in the endowment statement on Ar. 3334. That statement, however, was written at least eight years before the one penned by Hand no. 3, suggesting that a same source of vitriol was available on the market for a longer period of time and that perhaps other sources were available as well.

View Full Size

Figure 11.1

Comparison of the chemical composition of iron-gall and mixed carbon-iron-gall inks.

• Download Figure
• Download figure as PowerPoint slide

Based on the class identification and the semi-quantitative evaluation of the XRF data, it appears that none of the writings attributed to different hands share the same characteristics, discarding the hypothesis that the apposition of endowment statements and mottos was a collegial operation of people working together and sharing the same materials. The results seem, instead, to suggest a very personal ink usage, although we were able to analyse more than one note type penned by the same person on two different manuscripts only in the case of Hand no. 1. According to the abundance of this handwriting compared to the others, we provisionally identify Hand no. 1 as the handwriting of the first librarian. He used a glossy, deep black ink for both the endowment statement and a series of mottos on Ar. 3272 and the endowment statement on Ar. 3342. As an example, we show in fig. 11.3 the results of the motto on Ar. 3272, fol. 104^b, identifying the ink as carbon: there is no change of opacity and intensity in either the IR or UV images (fig. 11.3a–b), and only traces of iron and lead are detected with XRF (fig. 11.3c). Due to the aforementioned limitations, it is not possible to unequivocally confirm that the two endowment statements and the series of mottos were written with exactly the same ink, however, the scribe clearly wrote them with inks of the same class, the same optical characteristics and a very similar elemental composition. Additionally, the endowment statements were written within a three months interval, highly suggesting the ink is, indeed, the same.

View Full Size

Figure 11.2

a) Left, visible light image of Ar. 3315, fol. 1^a, with red boxes indicating the location of Dino-Lite micrographs in 2b and 2d, and right, IR light image of the same area using the LWP1510 filter; b) Dino-Lite micrographs from left to right in VIS, UV, and NIR light; c) XRF spectra of the ink and paper, analysed in the same area identified in 2b; d) Dino-Lite micrographs from left to right in VIS, UV, and NIR light.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Figure 11.3

a) Top, visible light image of Ar. 3272, fol. 104^b, with a red box indicating the location of Dino-Lite micrographs in 3b, and bottom, IR light image of the same area using the LWP1510 filter; b) Dino-Lite micrographs from left to right in VIS, UV, and NIR light; c) XRF spectra of the motto ink and paper, analysed in the same area identified in 3b.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Figure 11.4

a) Top, visible light image of Ar. 3334, fol. 1^a, with a red box indicating the location of Dino-Light micrographs in 4b, and bottom, IR light image of the same area using the LWP1510 filter; b) Dino-Lite micrographs from left to right in VIS, UV, and NIR light; c) XRF spectra of the endowment statement ink and paper, analysed in the same area identified in 4b.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Figure 11.5

a) Top, visible light image of Ar. 3342, fol. 1^b, with a red box indicating the location of Dino-Light micrographs in 5b, and bottom, IR light image of the same area using the LWP1510 filter; b) Dino-Lite micrographs from left to right in VIS, UV, and NIR light; c) XRF spectra of the motto ink and paper, analysed in the same area identified in 5b.

• Download Figure
• Download figure as PowerPoint slide

In 1197/1783, Hand no. 2 penned the endowment statement on Ar. 3334 with a mixed carbon-iron-gall ink (fig. 11.4), and a mixed carbon-iron-gall ink was also used for the three mottos written by Hand no. 10 on Ar. 3342 (fig. 11.5), although the ink has a different elemental ratio compared to that of Hand no. 2 (fig. 11. 1). We do not know if these mottos were penned in 1197/1783 at the same time as the endowment statement by Hand no. 1 or later. The abundance of ink classes and recipes detected, even during the same time span, reinforce the idea that different inks were available contemporaneously, either premade and sold on the market or prepared by the scribes themselves, and that the latter were using their own inks to write on al-Jazzār’s books.

Finally, all the inks used for the seals are carbon-based and, unsurprisingly, do not correspond to any of the inks used for the handwritten annotations. In some cases, it is possible to observe a small increase in the iron and lead peaks in the area around the seals compared to the paper, but the amounts are too low and heterogeneously distributed depending on the point of analysis to exclude the possibility that they may be merely impurities found in the writing support. Additionally, no clear correlation was noticed between the impurities in the inks and the seals stamped with the two matrices. It is, therefore, impossible to determine whether two different inks were used or if the matrices were made of different materials that left specific trace elements while stamping.

5 Conclusions

Overall, the observations suggest that a multitude of people had access to and responsibility over the books of the Jazzār library—challenging the assumption, based on the statement in the endowment deed, that only one position of librarian was envisaged for the complex. It is possible, for reasons unknown to us, that several people alternated in this position during the institution’s active years, or that users of the library and school had a more proactive role than we expect. It is also possible that some of the annotations were added during the revision of the library’s possessions that led to the production of the library inventory in 1215/1800.¹⁹ Extending the palaeographical analysis to other manuscripts from the collection might prove instrumental to defining each actor’s contribution to the annotation of the books. The application of the Handwriting Analysis Tool (HAT) developed by Hussein Mohammad (CSMC, Universität Hamburg) could be of help in the identification and attribution of hands.²⁰ Enlarging the pool of handwriting analysed with archaeometric methods could answer additional questions concerning scribal habits and preferences, and perhaps clarify the temporal distribution or the order of the annotations.

As expected, the inks employed for stamping do not correspond to the inks used for the annotations. It was not possible, however, to discriminate the carbon-based inks of the seals, or clearly define the presence of metallic elements in some of the annotations, as it is unclear whether the trace elements are truly present in the inks—an impurity from the production process of the ink, or a contamination from the metallic seal matrix or the inkwell could have occurred—or are due to the contribution of the writing supports. Such an open question can only be answered through the future application of XRF mapping.²¹