PREFACE

CONTENTS

I VERANTWOORDING EN CONCLUSIES

II JUSTIFICATION AND CONCLUSIONS

1.INTRODUCTION

2.AIMS OF THE SURVEYS

3.RESEARCH METHODS AND DEFINITIONS

4. THE DAMAGE SURVEY CONDUCTED AT THE GENERAL ARCHIVES OF THE NETHERLANDS (ARA) AND THE KONINKLIJKE BIBLIOTHEEK (KB)

I VERANTWOORDING EN CONCLUSIES

II JUSTIFICATION AND CONCLUSIONS

1. INTRODUCTION

Concern about the condition and preservation of archive and library material has made a quantum leap over the last decade; general public concern about the threat to the environment has contributed to this in no small way. Attention has concentrated on the problem of acidification in nineteenth and twentieth century collections in archives and libraries. Archivists, librarians and restorers have come to realise that measures are necessary if the autonomous deterioration of paper caused by the materials used in the paper manufacturing process after about 1830 is to be brought to a halt. Failure to take such measures will inevitably lead to the withdrawal of more and more archive and library material from use by the public.

Studies in other countries in the late seventies demonstrated that particularly in libraries as much as 30% or 40% of the books were brittle, in other words they could not be used without the paper crumbling. These figures were based primarily on estimates. A statistically valid survey was needed. The Stanford University Library was among the first to develop and publish a useable method (1979).

In the early eighties work in the Netherlands developed along two fronts. As a result of the report published in 1980 by the Institute for Social Research (IVA) of the Catholic University of Brabant at Tilburg, staff in record offices and libraries sought to answer the question of how to make a more accurate and reliable assessment of the real damage caused by acidified paper. On the other front, the Ministry of Welfare, Health and Cultural Affairs commenced a study in 1984 of the environmental impact on material in archives and libraries which had already suffered damage. This was known as the ACBAM study (the Dutch acronym for Damage to Cellulose-containing Material in Archives, Libraries and Museums). This study, the final stage of which will be completed in 1993, has in the meantime produced a publication setting out a number of preventive measures for protecting archive and library material from the influence of air pollution.

Growing concern for the preservation and restoration of acidified paper also led to an increase in political interest. Reports and recommendations from government advisory bodies have followed one another in quick succession in recent years. This broader-based interest in the subject resulted in January 1990 in the presentation to Parliament by the Minister of Welfare, Health and Cultural Affairs and the Minister of Education and Science of the policy memorandum entitled The Conservation of Paper Materials in Archives and Libraries (Tweede Kamer, 1989-1990, 21 443, nos 1-2). Among other things, the policy letter announced that the two ministries in question wished to institute a year-long project in this field, with a major role being assigned to the Koninklijke Bibliotheek (KB) and the General Archives of the Netherlands (ARA): in formal terms, the Public Records Department.

Within the framework of the project year, the main thrust of the endeavour would fall, in addition to research on the treatment of paper suffering the effects of acidification, on the conduct of damage surveys. It is clear that without unequivocal data on the dimensions of the problem, the search for solutions can only generate a further spate of unanswered questions.

On 30 October 1989 -during a preparatory seminar on damage surveys (see annex C)- the Koninklijke Bibliotheek and the General Archives announced the inception of a National Preservation Office (CNC). The CNC assumed responsibility for the mass conservation project year which commenced officially on 1 July 1990.

The first sub-project on the agenda was a damage survey of post-1800 paper material held by the Koninklijke Bibliotheek and the General Archives. At the request of the Ministry of Welfare, Health and Cultural Affairs, in its capacity of one of the initiators of the project year, an interim report on the damage survey was compiled in October 1989 by the CNC for use in preparing the policy document on the Endangered Cultural Heritage. The policy document was submitted to Parliament in December 1990. The CNC's interim report, which could only provide indicative data, is included in that document as annex 5.

The report gives the research results of the damage survey for the entire Koninklijke Bibliotheek and General Archives collections, together with an explanation of the research method employed.

2. AIMS OF THE SURVEY

From 30 October to 2 November 1989 a seminar was held in the Koninklijke Bibliotheek complex attended by experts from the Netherlands, Sweden (Uppsala University Library) and North America (Library of Congress). Its aim was to exchange views and experiences in the field of research on paper damage and following on from this to devise reliable research methods to tackle the problem in the Netherlands. On the basis of the work done at the seminar -the general conclusions are given in annex C- the CNC drew up four objectives for the damage survey.

a. Information must be obtained on the condition, and the quantities affected, of nineteenth and twentieth century paper in the Koninklijke Bibliotheek and General Archives, with special reference to the problem of damage to paper by acidification.

b. Information must be obtained on current conservation requirements and/or the backlogs in conservation work which have been identified and, if possible, data should be provided to enable a prognosis to be drawn up for the future conservation requirement.

c. An indication must be given as to the most appropriate measuring method for the selection of individual items eligible for conservation measures.

d. The survey results must be suitable for use in the broader field with Dutch archives and library collections and they must be capable of comparison with the results of international studies in this field.

The span of the survey was bounded by the years 1800 and 1980/90. The selection of 1800, and not 1830 or 1850, as a starting point was made on purely pragmatic grounds. For instance, the Koninklijke Bibliotheek has special regulations applicable to its pre-1800 collection which affect the placement of collections in the repositories. This has no other consequence for the damage survey than that of imposing a restriction on the time span covered. A damage survey of archive material dating from before 1800 will be carried out as a separate project by the Public Records Department.

The decision to conduct the damage survey at the General Archives and the Koninklijke Bibliotheek was based on the fact that, in view of the size, diversity and representative nature of their collections, these two institutions offer the best opportunity for extrapolating the results of the survey to other collections in the Netherlands. It might be expected that a sufficiently large, statistically independent random sampling from these collections would produce reliable conclusions. In addition to this, both institutions have close links in terms of policy-making and implementation with the conservation issue at national level.

However, the project proposal for the mass conservation year put forward as a reference framework the possibility of conducting a follow-up damage survey which, if the evaluation of the survey by the Koninklijke Bibliotheek and the General Archives found that it had produced insufficient data on methodology and research, could be carried out in a university library and a municipal records office.

3. RESEARCH METHODS AND DEFINITIONS

The preparatory seminar held towards the end of 1989 concluded that a statistical research model is essential to a reliable damage survey. In setting up the model recourse was had in large measure to the experience and knowledge of the staff -present at the seminar- of the Eindhoven-Kempenland Regional Record Office, where a damage survey of a number of special collections had been conducted earlier. In this way the scope of the survey at the General Archives and the Koninklijke Bibliotheek could also be expanded. Extensive discussions during and after the seminar led to the conclusion that a modified version of the Stanford and Library of Congress method would best meet the requirements deriving from the objectives of the damage survey at the General Archives and the Koninklijke Bibliotheek.

The terminology employed in the damage survey of the General Archives and the Koninklijke Bibliotheek corresponds with that of international surveys. This means that the study employed the definition for brittle paper stating that it must have a hand fold index of less than 6, and for weak paper, a hand fold index of between 6 and 20. From the earlier surveys conducted in the Netherlands at the Koninklijke Bibliotheek and the Eindhoven-Kempenland Regional Archives it was known that the number of ruptures diminishes substantially after 20 hand folds. Use also plays a role here: the actual physical handling of paper with a low hand fold index increases the risk to its survival. The two categories are jointly referred to in this report as 'endangered material'.

By contrast with international surveys, in the presentation of the results of the present survey (see part II) no relationship was sought with the material condition of the covers and binding. These data were registered but were not included in the correlation. It was the view of the CNC that the damage survey should establish primarily the quality of the paper and it was anticipated, partly on the grounds of the Swedish results, that including data on the material condition of the volumes would have too great an influence in negative terms on the final result. Swedish studies have moreover already demonstrated that the individual quality criteria overlap each other on this point. There are grey areas between the categories of 'cover' and 'binding' which would introduce an indeterminate margin of error in the final results.

The research method selected amounts in broad outline to taking a representative sample from the records and collections, determining a number of relevant physical and chemical properties of the paper, statistically processing these measurements and establishing correlations between the various groups of data. All the measurements were made by qualified restorers in the restoration workshops of the General Archives and the Koninklijke Bibliotheek. Dr J Kragten of the Analytical Chemistry Laboratory of the University of Amsterdam was commissioned, on the advice of the project leader for conservation at the Ministry of Welfare, Health and Cultural Affairs, to conduct the statistical research.

4. THE DAMAGE SURVEY CONDUCTED AT THE GENERAL ARCHIVES OF THE NETHERLANDS AND THE KONINKLIJKE BIBLIOTHEEK

4.1 POPULATIONS AND SIZE OF THE SAMPLE

The point of departure of the damage survey was that the results should be representative of the situation in the Netherlands. For this reason, a number of collections at the General Archives were excluded from the sample. These were archives which had not been created in the Netherlands, for instance those originating in the former colonies and overseas territories of the Netherlands or those inscribed on a special type of material such as charters (parchment), maps and drawings. (A separate damage survey of the latter category will commence in 1991.) Taking the floor plan of the repository of the General Archives as a basis, 10,731 racks covering a total of 63 km were included in the sample.

At the Koninklijke Bibliotheek the nineteenth and twentieth century collections were divided into four sub-collections: monographs, periodicals, newspapers and what is known as the Depot-collection (Repository of Dutch Publications). It was believed that the last mentioned in particular would produce important data for the prognosis. Two Mega items were taken as the starting point in establishing the total number of items in the Koninklijke Bibliotheek sample.

The size of a sample determines the accuracy with which the results of a survey can be presented. Considering the available information derived from other surveys and the required accuracy of a great number of parameters is decided, in consultation with a statistician, that a sampling ratio of 1 to 800 would be used. A figure was thus obtained for the number of items in the sample for each institution of 2500, excluding the reserve sample added to cover items which were temporarily absent (on loan, unavailable owing to their unusual form (e.g. a bibliophile edition), and so on). The reserve sample consisted of approximately 20% of the sample (500 items). It was decided that the sample at the General Archives should consist of 3000 items, including a 20% reserve sample. The number of items in the sample at the Koninklijke Bibliotheek should work out at 2800, including a 12% reserve sample. The difference in the reserve samples of the two institutions was caused by the number of empty shelves in each.

With the sampling ratio selected, the size of the sample (1 to 800) is such that for virtually all the results 95% reliability and an approximate margin of error of 2% could be obtained. It is possible to draw sufficiently reliable conclusions with this coincidental margin of error. It should be realised, however, that in order to gain double the reliability obtained in the present case, four times the number of measurements would have to be made.

4.2 SAMPLE DISTRIBUTION AND SAMPLING

The floor plans of the repositories and the racks they contain (General Archives) and the distribution of the storeroom and repository stacks over the various floors of the building (Koninklijke Bibliotheek) formed the basis for the distribution of the items sampled. These were used to gain an overview of the total number of stacks and the further distribution of the stacks within the four Koninklijke Bibliotheek sub-collections.

A closer inspection of the repositories at the General Archives showed that about 15% were empty. It was therefore decided, as already indicated, to increase the entire sample to 3000 items. These were distributed over 10,731 racks.

The four sub-collections at the Koninklijke Bibliotheek are spread over six floors. Often certain sections of two or three different sub-collections are housed together on one floor. An initial inspection showed that here, too, approximately 10-15% of the shelves were empty and the sample was therefore increased to 2800 items. Although it later became apparent that there was in fact more empty shelf space than had been estimated, no compensation for this was made in the sampling, and in fact it made no difference to the final results: when the sampling was done from the stacks concerned it was established that the entire Koninklijke Bibliotheek collection could be estimated at 1.5 Mega items. In order to achieve a proportional distribution of the items sampled among the individual sub-collections, the stacks in each collection were counted and the number and code of the sub-collections concerned displayed on the stacks. This resulted in the following sample distribution:

The total number of stacks was 10184.

All the items selected were drawn from the racks/stacks of the two institutions by means of a random method.

The racks at the General Archives, with very few exceptions, are 225 cm high. The number of shelves in each rack varies, but the width is exactly 100 cm. The selection of items was made as follows. A General Archives rack number was selected 3000 times from the series of numbers running from 1 to 10731 using a homogeneous distribution pattern (= random).

Virtually all the stacks at the Koninklijke Bibliotheek are also 225 cm high and 100 cm wide. The stack numbers were selected from the series of numbers in accordance with the number of items in the individual sub-collections, using a homogeneous distribution.

The rack or stack number obtained in this way was linked to an item number after ordering by the statistician. The position of the item to be sampled within a rack was determined using the coordinates 'height' and 'distance from the left'.

The height and distance were determined using a homogeneous distribution accurate to several centimetres between the limit values 0 to 225 and 0 to 100 cm respectively. The newspaper sub-collection was an exception; there the width of the stacks is 0 to 50 cm.

Height was measured from the floor (= 0 cm). If a distance as calculated fell between two shelves, the lower of the two was selected. If the height measured was above the rack/stack, the highest shelf was selected. If the distance measured fell below the lowest shelf (table between 0 and 8), the lowest shelf was selected. The box (General Archives) or volume (Koninklijke Bibliotheek) to be selected was determined by measuring from the left-hand upright. In doubtful cases the left-hand box or item was selected. The box or item indicated was removed in the position (vertical or horizontal) it was in at the time (see table 1)

At the Koninklijke Bibliotheek, a book was opened in a single movement by the restorer in order to find the centre. A strip of paper was then cut off the right-hand page.

Additional coding was required at the General Archives. If there was more than one file/folder/volume/part etc. in a box the centre file etc. with an uneven number was taken. Codes were used within a file etc, for instance: B9, E5, +M6 and -M12. B9 means the 9th sheet counting from the beginning, E5 is analogously the 5th sheet before the end. The centre of the file etc. was determined by 'file opening at the estimated centre' (i.e. opening with a single movement without making any correction): -M12 therefore means the 12th page before the centre, +M6 is the 6th page after the centre. The system was found to be quite straightforward (see table 2).

4.3 DATA TO BE MEASURED AND METHOD

The data sought were determined on the basis of the conclusions of the preparatory seminar held in 1989. In order to facilitate the procedure, registration data were printed on envelopes (see annexes A and B).

Registration was divided into two sections. General and random data were entered in the general section, while the results of the measurements obtained by the restorers were entered in the second section.

Originally, an investigation of the composition of the paper fibres was to be part of the measurements carried out by the restorer. This fibre analysis was considered necessary to demonstrate the distinction between groundwood paper and wood-free paper. It was thought that this, in addition to the registration of paper types (smooth, fibrous: see annexes A and B), would enable more to be learned about the different types of pulp used in manufacturing paper at various times. In order to learn more about how this should be done, the Koninklijke Bibliotheek's conservation scientist took part in an international seminar in England on identifying the composition of paper. The seminar revealed that the microscopic examination of various raw materials used in making paper required some degree of experience on the part of the researchers. It also became apparent that the preparation of the paper sample, the making of the material for microscopic analysis, applying the various dyes and the microscopic analysis itself also demanded a high degree of accuracy. However, the most important factor of all is that the interpretation of the fibre preparation is made more difficult by the circumstance that the fibres were usually damaged during manufacture and that, particularly in twentieth-century papers, use is often made of mixtures and combinations of chemical, mechanical and thermo-mechanical paper pulp. In the light of this information it was decided not to conduct a fibre analysis, partly also because the lignin test would also provide important information on this point.

With the aid of the sample lists prepared by the statistician, the items were drawn from the repositories by the restorers. After this the following measurements were carried out at the workshops by qualified restorers:

• Determination of the fold index

  The method used to conduct this test differed in the two institutions. At the General Archives only the hand fold index was used, while at the Koninklijke Bibliotheek machine folding was employed, in addition to hand folding of the following page. Since the margins of archival material are frequently written upon, it was not possible to remove a strip, as required by machine folding. A hand-folded 3 cm corner had to suffice. The hand fold index was determined according to the prescribed procedure: folding to and fro -to a maximum of 20 folds- applying light pressure until the break occurs. The same corner of paper was then used for the other tests, with the exception of the pH measurement which was carried out on the document itself. The MIT folding endurance meter was employed to determine the machine fold index; this was also used by the Library of Congress laboratory. The minimum standard tension of 1 kg, as used in the Arial paper industry, proved to be too great for damaged paper, so the tension was reduced to 0.5 kg. The strip of paper used for the test should be at least 1.5 cm wide and 20 cm long. The maximum number of folds was 50. In general the strip was taken from the right-hand side of the page. If the margin was not sufficiently wide, the strip was taken from the lower edge of the page. These strips were also used for the other tests.

• The lignin test

  The method used to demonstrate the presence of lignin was the standard colour test using phloroglucinol (composition: 1 gr phloroglucine, 50 ml ethanol, 25 ml hydrochloric acid). The colour was determined visually. The four categories (colourless (negative), brown, mauve, purple) were numbered from 0 to 3.

• The acidification test (pH)

  The method used here was surface measurement according to the TAPPI 529 standard. The tests were carried out on the sheets of paper (General Archives) or on the relevant strip of paper (Koninklijke Bibliotheek), from which the sample for the other tests was taken. The pH value for neutral paper was set at 7.

• The alum test

  The alum test was based on coloration formed with a aluminium-indicator, aluminon (composition: 1 part of aliminon in 1000 parts of distilled water). There were two possible reactions: no colour reaction (0) or positive (1). It should be mentioned however that aluminium possesses low reactivity and is highly dependent on the occurrence of a polyhydroxide complex reaction by the aluminium in the paper.

• Determining the thickness of the paper

  This test used a paper thickness meter (General Archives: Mitutoy 7001 DGFM, Koninklijke Bibliotheek: Frank 16874).

• The discoloration test

  Discoloration was established by visual observation. There were only two possible answers: not discoloured (0) and discoloured (1).

• Type of paper

  The type of paper was established by visual observation and manually. The categories (fibrous, smooth, copying paper, newspaper, photocopy, art paper, other) were numbered from 1 to 7.

The damage survey was primarily intended to produce reliable data on the current extent of deterioration of nineteenth and twentieth-century paper. In addition it is to be of vital importance that specific sub-problems could be identified by correlation of the data and, where possible, a prognosis made of the anticipated damage if no further conservation measures were to be taken. In order to achieve the latter a number of data were correlated statistically. The survey correlated the following data.

- decade

- discoloration

- type of paper

- lignin

- pH value

- alum

- hand fold index

- paper thickness.

For the Koninklijke Bibliotheek the country of origin (not deemed necessary at the General Archives) and the machine fold index were also entered as data for correlation.

The measurements to be correlated were coded as follows:

The results of this correlation were presented in both graph form and tables by the statistician. Because of the survey result of the total collections of the General Archives of the Netherlandsand the Koninklijke Bibliotheek an adjusted correlation scheme has been used for the separate sub-collections and a number of statistic deepenings has been carried out. Combinations of three parameters have been correlated here.

The statistical processing of the data obtained by the damage survey was carried out externally. The CNC called in Dr J. Kragten of the Analytical Chemistry Laboratory of the University of Amsterdam for this task. The coordination of the work of the statistician by the CNC was done by the project coordinators (from the General Archives and the Koninklijke Bibliotheek), the conservation scientist from the Koninklijke Bibliotheek and the head of the restoration workshop of the General Archives. The CNC project coordinators also held consultations on several occasions with the statistician who had earlier carried out a damage survey of the records in the Municipality of Eindhoven.

In order to be able to work as flexibly as possible with different types of data, for instance years which had to be converted to decades, discoloration and alum content which were only represented in binary code and the names of countries which were sometimes coded differently, processing by means of a spreadsheet programme on a PC appeared to be indicated. This permits more flexible processing than with SPSS and SAS on a mainframe. The disadvantage of this method is that programming in macros is more complex and more time-consuming, although the fact that the macro operations themselves can be followed on the screen is an advantage. It was therefore decided to use the LOTUS 123 spreadsheet program on an Olivetti 290 PC-AT with 2 Mb of RAM.

A special input program was written and compiled for input by hand of data from the data registration forms (see annexes A and B). The program checked the input of various types of data for correctness of the measurements and input. These checks ensured that dates were between 1799 and 1991; alum or discoloration only 0 or 1; type of paper only from 1 to 7; lignin only 0 to 3; pH values lower than 7.0 (very few samples have a pH higher than 7); hand fold index up to 20; machine fold index up to 50; and paper thickness in millimetres and less than 2.0 mm.

During data input (see tables 3 and 4) it was observed that data for some items were not available or were incomplete. This was due to the fact the restorers had been unable to obtain the data for practical reasons or had been unable to take measurements. An estimate of 10% empty fields was therefore included in the data matrix. There were very few incorrect measurements (fewer than 0.2%). Where possible the statistician corrected any errors in the input stage. Where this was not possible the data in question were entered as 'not present' (empty cell in the matrix).

The sub-collections at the Koninklijke Bibliotheek were entered separately because they were numbered differently. The files were later combined into one large Koninklijke Bibliotheek matrix with an additional code for the sub-collections. Using a special parsing program, the ASCII files of both collections were converted into work files for processing with LOTUS 123.

An overview of the data to be correlated and the relevant table was drawn up in close consultation with the CNC project coordinators (see table 5). These correspond to the survey results presented in chapter 7. According to this scheme, the figures 3-8 refer to a correlation between foxing and the hand fold index.

Preliminary information on correlations such as this was not available. The correlation experiments were first carried out at the General Archives because the composition of the records was homogeneous and they formed a single entity (no sub-collections).

The correlations were carried out as follows. The columns in the total files which were to be correlated were copied to a separate file. Then the rows containing empty cells were removed from the data matrix. In the correlations with number 1 of table 5 the years are grouped in decades; for instance, 1895 falls under the nineties (1890).

In the correlations with numbers 6, 8 and 9 in table 5, depending on the information sought and the result, the average was sometimes calculated with the interval within which 95% of all values fall (the confidence interval of 95%, indicated by the limits BI 95%+ and BI 95%) and sometimes the number of items within an interval was counted. See chapter 7 and part II for a visual presentation of the correlations.

The confidence intervals and the standard deviations at various points which are given as an indication of the distribution of the values measured concern individual observations. They show the dispersion of the individual observations round the true values. Owing to the limitations of the sample the average values are also subject to dispersion. However, this is smaller by a factor equal to the root of the number of items and is thus in most cases an order of magnitude smaller.

Where the results span an interval (e.g. in the pH measurements), the lower value is included but for statistical reasons the upper value is not included in the interval. A pH area of 4.5-5 therefore includes the value 4.5 but not 5.0: the value 5.0 occurs in the following interval.

6.1 STAFF DEPLOYMENT

The CNC project proposal sent to both institutions in May 1990 assumed the availability of two project coordinators from the General Archives/Koninklijke Bibliotheek (part-time), four restorers (General Archives/Koninklijke Bibliotheek) and an external restorer for the more time-consuming tests. The work of this group was planned to take five months. The statistical analysis and the input capacity for this were contracted out. Since the Ministry of Welfare, Health and Cultural Affairs had requested an interim report to provide information for the Policy Document on the Endangered Cultural Heritage which was to be presented to Parliament in December 1990, the relevant consultative committee of the CNC decided to increase the staff of the project by adding two restorers for each institution. The total contribution on staff (capitalised) by both institutions therefore rose from Dfl. 106,000 to Dfl. 181,000. The advantage of this was that the tests could be completed in three months instead of the planned five; this was also partly due to the fact that since the external restorer was not in fact required to carry out the fibre analysis, he could also work on the other tests. From 1 July to 30 September 1990 the General Archives invested a total of 1600 hours in the damage survey, of which 320 hours were worked by the project coordinator (incl. the preparatory stage of the tests). The Koninklijke Bibliotheek spent approximately 370 hours on preparations (including numbering the stacks). From 1 July to 30 September 1990 it put in 1200 hours on the damage survey, approximately 150 of these being put in by the project coordinator. This makes a total of approximately 1570 hours. The external restorer spent about 160 hours working on the damage survey and worked out the remainder of his contract doing research relevant to the mass conservation project year. The statistician worked for a total of 100 working days, spread over the whole year of the project, on the damage survey: 22 of these were spent on program development, 33 on the processing of correlations and 25 on the presentation of the statistical results and consultation on this with the project coordinators. Data input by a data typist hired by the statistician took 20 days.

6.2 FINANCIAL RESOURCES

From the point of view of efficiency, the CNC's financial accounting was done by the financial department of the Koninklijke Bibliotheek. However, both institutions remain responsible to their respective departments for the use of the funds at their disposal. In view of current financial regulations for both institutions, internal arrangements were agreed about accountability to the their departments.

The budget for the damage survey attached to the project proposal submitted by the CNC amounted to Dfl. 274,000, including its own personnel costs. The contribution by the General Archives and the Koninklijke Bibliotheek from their own funds was Dfl. 106,000 (staff costs), excluding the additional staff allocation. The remainder (Dfl. 186,000) was charged on a proportional basis to the Ministry of Welfare, Health and Cultural Affairs and the Ministry of Education and Science. From this amount the costs of the external statistical work and data processing (approx. Dfl. 124,000), the purchase of the MIT Folding Endurance Meter (approx. Dfl. 20,000) and a computer and peripherals for the data input, data processing and sampling administration (approx. Dfl. 17,000) were met. The remainder was used for organisational expenses such as the publication of the survey results and travelling expenses.

The graphs and tables belonging to this report are not available in this electronic version. However, the index of these has been added. The printed graphs and tables mentioned in this index may be viewed at the Koninklijke Bibliotheek.

The survey results in the printed version consist of three parts: a General Archives damage survey, a Koninklijke Bibliotheek damage survey, and a Koninklijke Bibliotheek sub-collections damage survey. For the parts of the survey that are related to the whole of the KB collection, the survey results are always presented in tables. In most cases these results are also presented in a graph. For the sub-collections the results of the survey are usually presented, unless the relevant data did not yield much information. For example, the correlations of the paper thickness-findings have sometimes been omitted, while in other cases presentation has been restricted to mere tabular mentioning.

(Original text of Par. II.7 as presented in the paper edition:)For reasons of clear layout and legibility the graphs and tables belonging to this report have been placed after the introduction. The order for this is Koninklijke Bibliotheek damage survey, General Archives damage survey and KB sub-collections damage survey. In those cases where relevant data have hardly yielded any information at all -for example the correlations of the paper thickness- findings have sometimes not been presented, while in other cases presentation has been restricted to a mere tabular mentioning.

1. 1. Taking as the starting point the aims, definitions and the method used for the damage survey, the following picture emerges for the General Archives and the Koninklijke Bibliotheek: For the period under review (1800-1990), at the General Archives 1.5% brittle paper was found and 6.3% of the paper was in the 'weak' category. Calculated on the basis of individual years, the boundary years could be set with greater precision at 1840-1950. Within this period 2.9% of paper was found to be brittle and 8.8% to be weak. The survey results showed that paper dating from 1870 to 1880 was in the worst state of deterioration: 7% brittle and 20% weak (see: ARA file 1-8). For the period under study (1800-1990), at the Koninklijke Bibliotheek there was 2.2% brittle paper found and 4.4% fell into the 'weak' category. On the basis of individual years, the boundary years could be set with greater precision at 1840-1950. Within this period 8.0% brittle paper and 14.8% weak paper was found. The survey results showed that paper dating from the 1880-1890 was in the worst state of deterioration: 21% brittle and 16% weak (see KB File 1-8). The sub-collections of the Koninklijke Bibliotheek are divided as follows: monographs: brittle 2.7% weak 12%(1840-1950); newspapers: brittle 24.3%, weak 18.7%(1840-1950); periodicals: brittle 2.6%, weak 14.7%(1860-1930). No brittle paper was found in the Depot-collection. The above figures differ favourably from the very provisional figures given in the CNC interim report for the Policy Document on the Endangered Cultural Heritage.
2. 2. In determining the short-term conservation requirement (within 10 years), the 'brittle' and 'weak' categories were combined (endangered material). As regards the General Archives, this means that 7.8% of the 63 km of the classified collection which represents the 1800-1900 period should be subject to conservation measures, i.e. more than 4900 metres. As far as the Koninklijke Bibliotheek is concerned, the same period, 6.6% of the 1.5 million volumes should be treated. This is equivalent to 99,000 volumes (3246 metres). With regard to the treatment of endangered material, the figures mentioned above give an indication of the extent of the material requiring treatment. As the conclusions set out below will show, selection to the level of the individual item (single document/volume) would be too labour intensive within the framework of mass conservation. For instance, it has been demonstrated that on the basis of the graph showing the hand fold index per decade (General Archives of the Netherlands), the endangered records are concentrated in the 1850-1940 period. Since selection of individual items is not possible, this means that the amount of material which should undergo conservation, determined by selection on the basis of decades, would work out at 42% of the collection, or 26.5 kilometres. At the Koninklijke Bibliotheek such an estimate works out in about a third of the total possession, 500.000 bindings.
3. 3. In preparing a prognosis it was necessary to gain some idea of the time frame over which the 'weak' category would deteriorate to the 'brittle' level. In principle the survey results, by quantifying the 'weak' category (hand fold index 6-20), supply a prognosis for the amount of material requiring attention, in addition to the 'brittle' category. The fact that material is becoming brittle is confirmed by the results of the test for the machine fold index. In view of the identical curves in the General Archives graphs of hand fold indices and the Koninklijke Bibliotheek machine fold indices, it may be assumed that the same also applies to the items subjected to the hand fold test. On the basis of the survey results no prognosis can be given for a time frame within which weak material becomes brittle. The rate of deterioration is determined by a whole complex of factors: variations in the material itself, storage conditions, frequency of use, climatological conditions, and so on - all variables which could not be included in the present survey. It is recommended that these aspects should receive special attention in the preparation by the Central Research Laboratory for Objects of Art and Science of a multi-year research programme dealing with paper.
4. 4. The survey results provide no grounds for setting up the method employed -selecting a random sample, followed by the statistical processing of the data- any differently from the way it was done in the present case. With regard to the statistical processing of data it should be observed here that, since this was done externally, the CNC is unable to draw any conclusions or make recommendations for changes.
5. 5. The conclusions about methods of selection for conservation treatment were based on the correlations between the hand fold index and the other paper parameters tested. This approach was chosen because at the General Archives only the hand fold index could be measured and because at the Koninklijke Bibliotheek this method was used in conjunction with the machine fold index. The survey demonstrated that the hand fold index is a reliable method. The term 'significant' as used here indicates that the differences shown extend beyond the confidence interval.
• Hand fold index / country (for the Koninklijke Bibliotheek only, see KB file 2-8): the differences between the percentages of brittle material per country are not significant. There is an average for all countries of 2.2%. The conclusion is that the country of origin is not a selection criterion.
• Hand fold index / discoloration (see KB/ARA file 3-8): there was a significant difference in both institutions with regard to the correlation between degree of discoloration and the hand fold index. The relationship was less marked for the General Archives than for the KB. At the General Archives 80% of brittle paper was discoloured; paper with a hand fold index of more than 20 showed approximately 55% discoloration. At the Koninklijke Bibliotheek 95% of brittle paper was discoloured; paper with a hand fold index of more than 20 demonstrated about 30% discoloration. The sub-collections of the KB are divided as follows: monographs: brittle 83% discoloured, hand fold index more than 20, 36% is discoloured; newspapers: brittle is 97% discoloured, hand fold index more than 20, 22% discoloured. From the Depot-collection 6.7% is discoloured all the material has a hand fold index more than 20). The conclusion is that, since at both the General Archives and the Koninklijke Bibliotheek paper with a hand fold index of more than 20 also shows a high per-centage of discoloration. Discoloration is not suitable as a selection criterion. The degree of discoloration can, however, be used as an indicator for priority conservation in the case of collections/items which have already been selected on other grounds for conservation treat-ment. Within such a group, discoloured material receives priority for treatment. For both the General Archives and the Koninklijke Bibliotheek the statistical analysis of the parameters discoloration/hand fold index/decade provided support for this conclusion (see ARA/KB deepenings 1-3-8).
• Hand fold index / type of paper (see KB/ARA file 4-8): the difference between the percentage of endangered material per type of paper is considerable. It is not surprising that newspaper had the lowest score: 43% of the Koninklijke Bibliotheek newspaper collection between 1840 and 1960 falls within the 'endangered' category: 24% has a hand fold index of less than 6 (brittle) and 19% has a hand fold index of 6 to 20 (weak) (see: KB sub-collection 'krant' file 1-8). As a selection criterion for the type of conservation method, the type of paper in the collections is a fairly good indicator as far as libraries are concerned, where the type of paper used can be traced relatively easily. For archives, where the variety of paper types within a file, and even within apparently similar types of documents (e.g. register series), is quite large, the type of paper is not suitable as a selection criterion.
• Hand fold index / lignin (see ARA\KB file 5-8): the graph shows that at the General Archives the percentage of endangered material among records on lignin-bearing paper is 6.3% and that for lignin-free paper 9.3%. At the Koninklijke Bibliotheek (KB) the percentages are 9.1% and 3.7% respectively. The sub-collections are divided as follows: monographs: of the lignin containing paper 9% is endangered, of the lignin free paper 6% is endangered; newspapers: of the lignin containing paper 16% is endangered, of the lignin free paper to little information is available; periodicals: of the lignin containing paper 9.5% is endangered, of the lignin free paper 1.85 is endangered. The conclusion must be that on the grounds of the data derived from the survey, the lignin test is not suitable as a selection criterion.
• Hand fold index / pH (see KB/ARA file 6-8 and KB sub-collection files, 1-8): the graphs of the test results at both the General Archives and the Koninklijke Bibliotheek demonstrate that the correlation between the average pH (5) and the hand fold index is not significant. It may be concluded that the pH value of paper is not a suitable criterion for conservation treatment. The statistical analysis of the parameters pH/hand fold index/decade did show that at the Koninklijke Bibliotheek the average pH for paper with a hand fold index of less than 6 is one pH unit lower than paper with a hand fold index of more than 6. Within the margin of error for individual measurements, however, this difference is not significant and therefore negligible (see KB Deepenings 1-6-8).
• Hand fold index / alum (see KB/ARA file 7-8): at neither the General Archives nor the Koninklijke Bibliotheek was there a correlation demonstrated between the alum content of the paper and the hand fold index. The doubts expressed at the beginning of the survey were therefore confirmed. It may be concluded that the alum test is not a suitable criterion for conservation treatment.
• Hand fold index / decade (see ARA\KB file 1-8 and KB sub-collections file 1-8): the graph is very clear: over the entire test period (1800-1990) the increase in paper of poor quality found between 1850 and 1940/50 is significant. The results obtained with the machine fold test run parallel to this. This method is quite suitable as a criterion for undertaking conservation measures, there being a preference for the hand fold method.

On the basis of the above, the following general conclusions may be drawn.

• The hand fold index correlations studied, with the exception of the 'decade' parameter, did not produce a generally applicable selection criterion for conservation measures of endangered books and documents.
• The instrument to be used for further refinements in a selected decade is the hand fold index.
• The presence of discoloration is suitable as a supplementary criterion for establishing priorities and, for libraries only, the determining of paper type.
• With regard to implementation, it should be realised that selection by means of the hand fold index, degree of discoloration and type of paper is labour intensive and often difficult to carry out, so that in fact the only remaining criterion is the 'decade' parameter. The consequence of this is that the quantity of material requiring conservation treatment may far exceed the theoretical conservation requirement. Depending on this factor, it should perhaps be recommended that an additional sample of smaller size should be taken. This would of course depend on the resources available.
• The tests conducted at the General Archives and the Koninklijke Bibliotheek and their correlations provide a good understanding of the role of the individual factors contributing to the deterioration of paper. It became clear, for instance, that differences in degree of deterioration are clearly related to discoloration and type of paper. The relationship was less clear in the case of pH value and the lignin content.
6. 6. The starting point of the survey at the General Archives and the Koninklijke Bibliotheek was that it must be possible to compare the results with research done elsewhere, notably the studies conducted by the Library of Congress, the university libraries of Yale, Stanford and Uppsala and, where relevant, the pilot study done at the Koninklijke Bibliotheek and the damage survey conducted at the Eindhoven-Kempenland Regional Archives.

   Before going any further, it should be mentioned that the published damage surveys carried out abroad give no detailed data on individual parameters. This makes it impossible to compare the results of the General Archives/Koninklijke Bibliotheek Survey down to this level.

   With regard to the period of 'poor-quality' paper, the ARA/KB results correspond to the Swedish studies and that done at Yale: their boundary years are also 1840 and 1950. Within this time frame, the occurrence of the poorest paper quality is concentrated in the years 1870-1910 (ARA 1870-1880 and KB 1880-1890). The comparison is less easy to make with respect to the Library of Congress since the bulk of the material tested (89%) there is no more than 80 years old. It will be clear that as far as the period being studied is concerned, these results will give a picture differing from that obtained from studies of material spanning a period of more than a century and a half.

   The results obtained from the machine fold test (Koninklijke Bibliotheek only) do agree in large measure with those of the Library of Congress, however. The General Archives/Koninklijke Bibliotheek survey demonstrates that the results of pH measurements do not correlate significantly with paper quality. A slight correlation may nevertheless be assumed: endangered paper material has on average a lower pH than 'good-quality' paper. The Swedish study and that carried out at Eindhoven-Kempenland confirm this: there, too, a slight correlation was found. The average lowest pH found by them was 4.5 (Sweden) and 4.2 (Eindhoven) and the average highest pH was 5.1 (Sweden) and 6.2 (Eindhoven). For the General Archives these figures are 4.3 and 5.6 respectively and for the Koninklijke Bibliotheek 4.5 and 5.6 (6.5 for recent paper). The percentages of brittle and weak paper, together termed 'endangered material', obtained from the ARA/KB survey cannot be readily compared. As was mentioned in the introduction, the ARA/KB survey focused on the quality of the paper and not on the material condition (binding, cover), which was included in the Yale, Stanford and Uppsala studies. Their percentages are thus considerably lower than those obtained from the General Archives and Koninklijke Bibliotheek survey.

7. 7. One of the aims of the ARA/KB survey was to ensure that the items selected for the sample should contain material representative of Dutch record offices and libraries. In this connection it is an indisputable fact that paper used for books is more homogeneous than paper used for public records. On the basis of the general data obtained from the data registration for the General Archives/Koninklijke Bibliotheek survey and from the various graphs, it has emerged that the sample material contained a large number of different types of paper of very diverse origin.

   Moreover, a major indicator for the representative nature of the sample was the virtually identical results obtained from both the General Archives and the Koninklijke Bibliotheek in respect of one of the most important parameters of the survey, the hand fold index. It may therefore be concluded that the sample is broadly representative of archive and library material held in the Netherlands. This leads the CNC to believe that a supplementary damage survey would add little new information of any significance to what has been obtained from the present survey and through the method employed.

8. 8. The present report has set out the general results of the ARA/KB survey. In addition to this the results of the damage surveys of the various sub-collections at the Koninklijke Bibliotheek have also been presented. The general results and the results for the sub-collections support each other. An additional advantage of presenting data for separate sub-collections is that priorities in selecting candidate areas for conservation treatment can be assessed with greater accuracy. Besides, findings of the sub-collections indicate that the overall priority should go to the newspaper collection in the first place and to the monographs in the second place. The periodicals have lowest priority while the Depot-collection has no priority at all, so far.
9. 9. The data registration forms prepared for the survey do not need to be altered.
10. 10. In carrying out the lignin test (phloroglucinol test) it was revealed that approximately 50% of the test material showed a coloration (brown) deviating from the standard colours (mauve/purple). An exploratory test carried out by the Netherlands Organisation for Applied Scientific Research (TNO) demonstrated that the test material which turned brown was negative (free of lignin). This fact was adopted for the purposes of conclusion, although the background to the phenomenon is not clear. The CNC therefore recommends that the reasons for this phenomenon should be included as a subject for study (by means of a series of FTIR measurements) in the multi-year research programme to be initiated in the course of 1991 by the Central Research Laboratory for Objects of Art and Science.
11. 11. As was mentioned earlier in the conclusions, although the parameters used for this survey do chart the extent deterioration of paper (brittle/weak) they do not shed much light on its causes. In the view of the CNC, research into the causes of deterioration should play a key role in the Central Research Laboratory's multi-year research programme. An example of such a study would be an investigation into the effects of metals contained in paper on oxidative decomposition. It is important that the results of the ACBAM (Damage to Cellulose-containing Material in Archives, Museums, Libraries and Museums)/STEP study, which is being implemented by TNO and of which the present survey is a part, should be made available to record offices and libraries in the Netherlands.