Visual inspection
The VC papers were visually examined, the dimensions of the papers were measured, and the dimensions and orientation of the paper mould were determined (Tab. 4). Photographs were taken in transmitted light to define and measure mould laid and chain line spacing and watermarks. The chain line spacing ranged from 22-32 mm, while the density of the laid lines ranged from 7/1 cm to 10/1 cm. Most of the watermarks are a coat of arms, two of them identify the origin of the paper mill for historical hand-made papers, while the others were assigned a label according to the IPH standard [40]. The dimensions of the paper mould were measured on HP samples, chain line spacing was in the range of 22–32 mm, and laid lines were in the range of 7–10 lines/1 cm. Six watermarks were found, and two were identified. One watermark from VC depicting the coat of arms of Carinthia was identified as a watermark from St. Ruprecht paper mill near Klagenfurt in Carinthia (Fig. 1a) [41] (present-day Austria) from the 17th century. The other sample is from a group of HP samples from the paper mill of Valentino Galvani from Pordenone (Fig. 1b) [42] (present-day Italy) from the 19th century.
Table 4 Papers from VC: dimensions of papers, watermarks, dimensions and orientation of mould
|
Sample group mark
|
Dating
|
Paper width x height /mm
|
Place of origin or use of paper
|
Appearance and type of watermark
|
Chain lines
|
Laid lines
|
|
1ND
|
1662
|
410-420x320
|
Zagreb
|
1 type of watermark Bishop’s crosier IPH: R1/1 or R1/2
|
Chain lines vertical
20-21 mm
|
Laid lines 9, 10 and 12/1 cm
|
|
2ND
|
1679
|
278x190
|
Bogenšperk
|
4 types of watermarks - coats of arms except one, endleaf watermark depicting the coat of arms of Carinthia
|
Chain lines horizontal
22, 23, 28 mm
|
Laid lines 9 and 10/1 cm
|
|
3ND
|
Prior to 1681
|
278x179
|
Bogenšperk
|
5 types of watermarks - coats of arms, endleaf watermark depicting the coat of arms of Carinthia
|
Chain lines horizontal
22-28 mm
|
Laid lines 7, 8, 9, 10/1 cm
|
|
4ND
|
1681
|
286x180
|
Bogenšperk
|
4 types of watermarks - coats of arms, endleaf watermark depicting the coat of arms of Carinthia
|
Chain lines horizontal
22, 23, 24 mm
|
Laid lines 9 and 10/1 cm
|
|
5ND
|
1685
|
360-370x420-440
|
Bogenšperk
|
18 types of watermarks; 12 coats of arms, 2 scales, 2 horns, 2 lions 1, Bishop's crosier IPH R1/1; covering sheets and endleaves coat of arms depicting the coat of arms of Carinthia
|
Chain lines horizontal, other vertical 16-28 mm
|
Laid lines 8, 9, 10, 11, 12/1 cm
|
|
6ND
|
1688
|
245 and 440x368
|
Nüremberg
|
2 types of watermarks - coats of arms and Hermes cross on 7 sheets IPH S3
|
Chain lines on the covering sheet and endleaf, horizontal. Book block sheets vertical; 28 and 29 mm
|
Laid lines 10 and 11/1 cm
|
|
7ND
|
1689
|
240x373-376
|
Nüremberg
|
3 types of watermarks - coats of arms
|
Chain lines horizontal
23-28 mm
|
Laid lines 8, 9, 10/1 cm
|
By analysing the dimensions of the papers, the imprint of the paper mould, and the position of the watermark, it can be concluded that, in all the researched books, the papers from VC originate from two paper mould formats. The dimensions of the smaller format are 245-278 x 360-380 mm and those of the larger are 360-370 x 420-440 mm.
The above examinations by visual inspection led to several conclusions. The same watermarks on different sheets do not have the same dimensions, which indicates that the manufacturer used several paper moulds for the same type of paper. It has been observed in several examples (3ND, 4ND, and 5ND) that the endpapers in front and back of the book were made from the same type of watermarked paper (coat of arms of Carinthia), while the leaves or folios of the book block were from different types with different watermarks. By examining the watermarks, it can be concluded that in most cases, they are coats of arms. From 144 research papers, watermarks were found on 38, of which 29 were coats of arms.
Microscopy analysis
The surface of the paper was analysed with a digital microscope on both VC papers and HP samples, while fibres were analysed by an optical microscope only for HP samples. Inspection of papers using a digital microscope revealed some specific fibres that appear to stand out from the majority of white fibres, as shown in Fig. 2. Several types of such fibres have been detected on both sets of samples, VC and HP. Thus, blue, red, light brown, dark brown, and black-brown fibres appear in all samples examined. By measuring the width of the fibres, it was observed that there are two types of fibres, thin (<150 µm) and thick (>150 µm), as shown in Fig. 3. Among the HP samples, 2D stands out, where the largest amount of different types of fibres was detected, as seen in Fig. 2. In samples 3D, 4D, 6D and 8D, thick light brown fibres were found, which could indicate that these are straw fibres. Thin fibres are present in samples 5D, 7D and 10D.
From the non-destructive microscopic imaging of the raw historical papers of Valvasor's collection (VC), including the outstanding items as fragments (HPs), it can be concluded that the most specific fibres that appear can be described as thin light brown fibres. The fibre is less than 150 µm thick, and its colour is light brown. Blue and red fibres were found in many papers and, according to width, can be characterized as thin fibres (Fig. 3a, 3b).
In addition, a thick light brown fibre was observed, which, in some places, has a width of over 400 µm and looks like straw fibres (Fig. 3d, 4a). In some places, thicker fibres were found deep in the structure of the paper (which made it difficult to measure them accurately) so that the white fibres were clearly visible on them, and it was possible to measure their width (Fig. 4). The width values of white fibres ranged from 11 µm to 29 µm, which would probably refer to two types of fibres specific to the 17th century: hemp and flax [29].
Flax and hemp fibres and, in some cases, cotton fibres are visible under the optical microscope (Fig. 5a). The flax fibre cells appear as long transparent, cylindrical tubes that may be smooth or striated lengthwise. The width of the fibre may vary several times along its length. There are swellings or nodes at many points, and the fibres show characteristic cross-markings. The fibre cell has a lumen or canal running through the centre; the lumen is narrow but clearly defined and regular in width [30]. The individual cells of hemp are, on average, 13 – 26 mm long. They are cylindrical in shape, with joints, cracks, swellings and other irregularities on the surface. The central canal or lumen is broader than that of flax, and the ends of the cells are blunt [29]. The fibres taken from HP samples were subjected to the Herzberg stain test for identification purposes [28,38]. Most of the analysed fibres were coloured reddish grey (Fig. 5a), which confirms that they are hemp and flax fibres, referred to in the literature as rags or fibres from old rags [3].
Thickness
The thickness of paper is one of its basic properties. The measurements of the thickness on the paper in VC were limited to the area along the edges of the papers and the area without a plate imprint. The thickness of the papers varied between 0.1-0.2 mm, while there were some papers in only two sample groups with thicknesses below 0.1 mm (2ND and 5ND) and few papers thicker than 0.2 mm. Values over 0.5 mm were determined on a composite sample, with paper glued on the textile base material. It was also noted that, within the same book, a small difference in thickness was observed between the endpapers and papers within the book block.
The measured values of paper thickness in six samples of HP range from 0.16 to 0.2 mm, whereas, for the other four samples, they are greater than 0.2 mm.
ANOVA showed significantly different values for the thickness of papers in VC and HP samples (p = 0.0077), suggesting that papers in VC, which are thinner, differ from other historical papers analysed.
Surface characterization
pH
The acidity/alkalinity of the paper is key information to preserve and protect the paper from degradation. Therefore, the determination of acidity/alkalinity a common method for conservation treatment on paper.
A method according to the TAPPI T529 [32] standard was applied only to HP samples, which prescribes measurements by applying a drop of distilled water to an appropriate place and measuring the pH in that drop. Most of the measured values are in the neutral or slightly acidic range, between 6 and 7. Three samples, 2D, 3D and 9D, stand out because their values are less than 6, and one of them, i.e., the 3D sample, has a pH value of 5. Values slightly above 7 were measured in two samples (1D and 6D).
Measurement of pH on the surface of HP samples revealed some destructiveness of the method due to damage caused by water. The stains occurred because of the water soluble components if paper or/and inks together with dust and dirt incorporated deep into the historical paper during a time, leaving tidelines. Such stains can be in some specific cases removed by conservation-restoration procedures, but because VC papers were not intended for further procedures, pH measurements were not applied to VC papers.
Optical Properties
To complement the visual evaluation of papers with the analytical method, the optical properties were measured. Brightness, yellowness, opacity and gloss were determined for VC and HP samples. These non-destructive measurements help to characterize papers. As optical properties depend on the structure of the paper, they can be useful for comparison with other historic papers and help to create a conservation-restoration protocol.
The brightness of papers in VC is mainly distributed in two groups, in the range between 40 and 50 % (41 papers) and between 50 and 60 % (37 papers). Most VC papers have a degree of yellowness in the range between 20 and 30 %. The measured opacity values are very high, between 90 and 100 %. Very low gloss values, below 2.5, tell us that the papers are uncoated. The most representative values for VC papers are between 1 and 1.5 in 77 % of cases (Fig. 6).
The brightness values of six samples of historical papers were in the range between 40 and 66 % (Fig. 7). For other samples, values below 40% were obtained, the lowest being 18.2 % for a HP sample from the 16th century. Seven samples showed yellowness values between 20 and 40 %, where the three samples from the 16th and 16-17th centuries had a darker colour, with yellowness values over 45 %. Very high values of opacity, over 90 %, were determined in HP samples, which can be classified into two groups: three samples were in the range between 90-95 %, and the remaining samples were in the range between 95 and 100 %. The measured gloss values on HP were approximately 1; with the exception of papers from the 19th century (sample 2D) with a measured gloss value of 0.73.
The measurements of optical properties of VC and HP samples revealed low brightness, high yellowness, very high opacity and extremely low gloss. Among the optical properties, a significant difference was observed in the case of gloss (p = 0.0001) and opacity (p = 0.0152), whereas brightness with p = 0.3268 and yellowness with p = 0.1365 showed no significant difference in values between VC and HP samples.
The Pearson correlation coefficient was determined along with the optical properties, thickness, pH content and water absorption of HP samples (Fig. 8). A negligible correlation was obtained between optical properties, water absorption and thickness, with the exception of opacity, for which a high positive correlation (r = 0.87) was obtained. A weak linear relationship between the calcium content present in papers and their optical properties was observed. With increasing amounts of calcium in the paper, the brightness and gloss increase, and the yellowness and opacity decrease. A strong linear relationship between all optical properties and paper pH values was observed (r > ±0.6). Papers with higher pH values have higher brightness and gloss and lower yellowness and opacity. Additionally, a strong positive linear correlation (r = 0.71) between the amount of calcium in paper and its pH value was determined.
A multiple regression analysis was used to determine whether there is a statistically significant relationship between properties and how two independent properties, thickness and pH value, influence the optical properties of HP samples. The results are summarized in Table 5. A multiple regression coefficient shows a strong relationship between properties. On the other hand, R2 and adjusted R2 show an acceptable predicted outcome only for opacity. A p-value below 0.05 for brightness and opacity means that a significant relationship exists between thickness and pH value, whereas no significant correlation was obtained for yellowness and gloss. The p-values for each independent property show that a highly significant relationship exists between the thickness and opacity (p = 0.009). The pH value showed a significant relationship with brightness (p = 0.015) and yellowness (p = 0.025) and marginal significance with gloss (p = 0.08).
Table 5 Results of multiple regression analysis for HP samples: multiple regression coefficient (R), coefficient of determination (R2), adjusted R2, p-values
|
|
Brightness
|
Yellowness
|
Opacity
|
Gloss
|
|
Multiple R
|
0.78
|
0.74
|
0.89
|
0.63
|
|
R2
|
0.61
|
0.55
|
0.80
|
0.40
|
|
Adjusted R2
|
0.50
|
0.41
|
0.74
|
0.23
|
|
p-value
|
0.037
|
0.063
|
0.004
|
0.169
|
|
p-value - thickness
|
0.299
|
0.323
|
0.009
|
0.582
|
|
p-value - pH
|
0.015
|
0.025
|
0.256
|
0.08
|
Spot Tests
The water absorption test is a micro-destructive method, as shown by the residual line spots formed after the measurement (Fig. 9). For this reason, the test was applied only on HP samples. Papers showed different behaviours regarding water absorption.
Thus, the least absorbent samples were 5D and 6D, which took between 6 and 7 minutes to absorb a drop of water, followed by four samples: 3D, 4D, 9D, and 10D, which took approximately 3 minutes to absorb a drop of water. It took approximately 2 minutes for samples 7D and 8D to absorb water, while samples 1D and 2D took less than half a minute. The absorbency of paper depends on the structure and surface properties of the paper and is primarily determined by the sizing.
A test to prove the presence of starch was carried out to determine whether the paper samples contained starch that was used for sizing in the production of handmade papers. In all ten tested HP samples (Fig. 10), none matched the expected results for the presence of starch in paper [43].
The aim of determining the presence of lignin in papers from the 16th - 19th century was to obtain indications if papers included fibres with more lignin. The test is primarily designed to determine the presence of wood fibres in paper, as wood contains a high proportion of lignin [43].
If the paper has a small amount of lignin-containing fibres, the individual fibres turn red and can be seen with the naked eye [43]. The spot test for determining lignin with phloroglucinol (Fig. 10) showed the presence of fibres containing lignin in most HP samples. If we obtain 2-3 lignin fibres on a 1 cm2 paper sample, this is evidence for a lignin content of less than <5%. Ten tested samples showed four categories of results regarding the amount of stained fibres in the paper. In only one sample, 10D (16th century), no stained fibres were found, whereas in two samples, 2D (19th century) and 6D (18th century), larger numbers of stained fibres were present (lignin content >5%). Two samples, 4D (16-17th century) and 7D (19th century), contained a moderate amount of stained fibres, and five samples (lignin content ≈5%), 1D (17th century), 3D (16-17th century), 5D (18th century), 8D (19th century), and 9D (16-17th century), contained small numbers of stained fibres (lignin content <5%). It can be concluded that all papers but one contain stained fibres and that a majority of samples contain a small number of stained fibres.
Microscopic and spectroscopic analyses
SEM-EDS
SEM-EDS analysis was applied to characterize the surface structure of paper and to identify fibres and elemental composition. Because this method is destructive, only the second set of samples, HP, were examined. SEM images of the paper surface and spectra with elements found in the recorded location are presented in the Supplement.
According to the EDS results, in most samples, O (oxygen), C (carbon) and Ca (calcium) were detected, except in samples 3D, 4D and 9D, in which Ca was not observed. Si (silicon), Cu (copper), Al (aluminium) and Ni (nickel) were detected in most samples as trace elements. Mg (magnesium) was also visible in trace quantities in samples 7D, 8D, 9D and 10D, while S (sulfur), Ag (silver), and Au (gold) appear in one sample (4D), which may represent contamination.
The SEM micrographs show differences between the samples; in some samples (5D and 8D), the spaces between the fibres are filled (Fig. 11a), indicating that surface sizing (Fig. 11b) was used during paper production and remained in some places until today. In other samples (1D and 10D), a larger number of unidentified particles are visible on the surface of the fibres.
XRF
XRF analysis (Supplement) was used to characterize the composition of HP samples and to complement the results obtained by SEM-EDS analysis. In all samples, calcium (Ca) was present, with the highest concentration/content determined in samples 5D, 6D and 8D, followed by 1D and 10D. In four samples (2D, 3D, 9D and 10D), iron (Fe) was also detected in addition to calcium. Trace amounts of silicon (Si), manganese (Mn), sulfur (S) were observed in all samples, potassium (K) was observed in seven samples, while aluminium (Al) was observed only in samples 3D and 4D.
In the SEM-EDS and XRF analyses, many elements have been recorded that can be related to the composition of the paper. Thus, sulfur (S), potassium (K), and in some samples, iron (Fe) were observed in the XRF spectra, while aluminium (Al) and copper (Cu) were observed in the SEM-EDS spectra. Sulfur is part of the molecular structure of gypsum (CaSO4 x H2O) and alum (KAl(SO4)2 x 12H2O); aluminium (Al) and potassium (K) are also present. Iron (Fe) was observed in trace quantities in all XRF spectra and can be attributed to the composition of iron alum (FeAl(SO4)2 x 12H2O). Silicon (Si) was detected in both elemental analyses and, according to the literature, is associated with the presence of straw. Straw stalks contain silicon oxide (SiO2) at 4-7% for wheat straw [29, 44]. The presence of most of the elements recorded in trace quantities during elemental analyses could be explained, while some of them, such as manganese (Mn), magnesium (Mg), titanium (Ti), copper (Cu) and nickel (Ni), could be attributed to sample contamination.
FTIR-ATR
FTIR spectroscopy, a simple technique for rapidly obtaining information about the chemical structure and crystallinity of cellulose samples, is often used in research on historical papers. We used this technique to record ATR-FTIR spectra on HP samples. According to the literature on the composition of historical papers, the expected components are cellulose fibres, including straw fibres [44], gelatine, gypsum, and calcium carbonate [18, 39, 45]. The raw materials used in papermaking in Europe until the 19th century were rags from used textiles (flax, hemp, cotton). The reference spectra of cellulose fibres have very similar absorption bands in the wavenumber range of 3660 - 2900 cm−1 (the stretching vibrations of OH and CH bonds in polysaccharides), with the band at 3331 cm−1 is characteristic of the stretching vibration of OH groups [46]. Typical bands assigned to cellulose in the range of 1630 - 900 cm−1, according to [46, 47, 48], are as follows:
- band at 1633 cm−1 corresponds to the vibration of water absorbed in cellulose,
- band at 1430 cm-1 corresponds to CH2 vibrations, HCH and OCH in-plane bending
- band at 1367 cm-1 corresponds to COH and HCC vibrations,
- band at 1335 cm-1 corresponds to OH and CH2 vibrations,
- band at 1155 cm-1 corresponds to COC asymmetric vibration,
- band at 1110 cm-1 corresponds to the asymmetric vibration of a glycosidic ring,
- band at 895 cm-1 corresponds to COC vibrations of glycoside bonds.
By examining the spectra of the individual samples and comparing them with the reference spectra, it can be noted that cellulose has two dominant regions: one is in the fingerprint region with a sharp band at 1029 cm-1 with two smaller bands at approximately 1050 cm-1 and 1100 cm-1, and the second region is in the area of stretching single bonds at approximately 3600-3000 cm-1. Thus, HP samples showed characteristic FTIR spectral features of cellulose. Furthermore, in most samples, absorption bands at 2900 and 2849 cm-1, attributed to CH stretching vibrations of cellulose and symmetric CH2 stretching vibrations of noncellulose polysaccharides, were observed as peaks or as shoulders. The absorption band at 2918 cm-1 attributed to asymmetric CH2 stretching vibrations in noncellulose polysaccharides was present as a sharp band in samples 5D, 6D, 7D, 8D, 9D and 10D, whereas, in four other samples , it was present as a shoulder (Fig. 12). This triplet is indicative of flax fibres [49]. In the study of Kostadinovska et al. [18], eight patterns were proposed to determine the prevailing type and quantity of fibres in the sample. By comparing the spectra of HP samples with the proposed patterns in all samples, the flax fibres were confirmed as the dominant component. In samples 2D and 3D, in addition to flax, a larger amount of hemp was identified, whereas, in sample 4D, a larger amount of cotton was present. Sample 1D showing absorption bands at 2902, 1506, 1031, 1000 and 898 cm-1 and shoulders at 2900 and 2851 cm-1 suggested that, in addition to flax fibres, hemp prevailed over cotton (Fig. 13). A weak absorption band at 1735 cm-1 present in all HP samples is characteristic of the C=O ester band in pectins, confirming the presence of hemp, although it could also show the presence of carbonyl groups of oxycelluloses found in degraded materials [47].
The absence of an absorption band at 1595 cm-1 and a very weak absorption band at 1505 cm-1, which are characteristic bands for lignin, confirm that all examined papers were made from pulped rags.
The spectra of additives (sizing agents and fillers) in paper, such as gelatine, starch, calcium carbonate and gypsum, overlap with the spectra of cellulose in the fingerprint region, so it is difficult to confirm with certainty the presence of individual materials. The absorption bands related to proteins below 1400 cm-1 could be assigned to amide III, those in the region of 100 to 1500 cm-1 to amides I and II, and bands over 3000 cm-1 (characteristic for NH and CH stretching) to amides A and B [50]. For papers sized with animal gelatine, the absorption bands at approximately 1645/1650 cm-1, associated with the C=O stretching vibration, and 1545/1550 cm-1, associated with the NH in-plane bending vibration, were observed in the spectra. In all HP samples, doublets at 1648 and 1645 cm-1 and weak absorption bands at 1550/1548 cm-1 were found, which could indicate the presence of gelatine in papers (Fig. 12).
Calcium carbonate shows a strong absorption band in the fingerprint region at approximately 1425 cm-1 and two weaker absorption bands at 875 and 712 cm-1 [51]. The presence of an absorption band at 1425 cm-1 in all HP samples, in addition to indicating the presence of calcium carbonate, is assigned to CH2 and CH vibrations in cellulose. The differences among samples were observed in the other two bands. The absorption band at 875 cm-1 was clearly observed in samples 5D, 6D and 8D and present as small peaks in samples 1D, 3D, 4D, 7D and 10D. In samples 2D and 9D, it was only a shoulder. A very weak absorption band at 711 cm-1 was observed in samples 5D and 6D and as a shoulder in samples 3D, 7D, 8D and 10D. Calcium in historic papers could be present in the form of calcium stearate, which shows an absorption peak at 2851 cm-1 and carboxylate bands at 1577 and 1541 cm-1 [12]. Both absorption bands (1575 and 1542 cm-1) were present in all samples, except at 2D, where only a band at 1542 cm-1 was observed.
Gypsum has two dominant sharp bands, one in the fingerprint region and the other in the region of asymmetric and symmetric OH single stretching bands, at 3394 cm-1. The other two bands typical for gypsum are at 667 and 595 cm-1[52]. In both regions, the bands overlap with the cellulose bands, making them difficult to identify. Nevertheless, we identified a band at 3394 cm-1 present as a shoulder in all samples and in samples 9D and 10D as a small peak. The absorption band at approximately 660 cm-1 (SO42- bending) overlaps with the OH out-of-plane bending vibration in cellulose and was detected at 667 cm-1 in all HP samples. The absorption band at 595 cm-1 was present as a small intensity peak in most samples and as a shoulder in samples 2D, 5D, 7D and 9D.
By comparing all recorded ATR-FTIR spectra, it can be concluded that there is a very small difference between the samples. As pulped rags were used for paper production, the presence of flax, hemp and cotton fibres in the samples is expected, with flax as the prevailing fibre component in most samples. In samples 1D, 2D and 3D, a larger amount of hemp was identified, and in sample 4D, cotton was also identified. In Fig. 12, the ART-FTIR spectra of three samples (1D, 6D, and 8D) from the 17th, 18th and 19th centuries are shown. The peaks mostly overlap, though in some parts of the spectrum they are different, which indicates the difference in the share of additives in the paper samples from the 17th to the 19th centuries. We can assume that gelatine was used for sizing in all cases. The presence of calcium compounds in the form of carbonate, sulfate or stearate could be predicted in all HP samples, though in different forms and quantities. Although the number of samples examined is small (10 samples), it is evident that there is a difference in the proportion of additives, which is confirmed by the research of Barrow [9], who concludes that over time, the amounts of additives (sizing and fillers) in the paper changed, which affected the durability and stability of the paper.