NATIONAL ARCHIVES AND RECORDS ADMINISTRATION

Preservation Information Paper No. 1
March 24, 1993

INITIAL FINDINGS: A STUDY OF SHRINK WRAPPED, SIMULATED BOUND VOLUMES

Background

Shrink wrapping of bound volumes as a technique in moving archival collections: The National Archives Experience

As anyone who has large collections of 18th and 19th century records can attest, bound volumes can represent a significant problem in an archival collection. The volumes used to produce fair copies of letters sent, registers, and indexes tended to be very large and often quite heavy. Incoming correspondence was often bound together regardless of the size of the paper. The precursor of carbon paper, the letter press book, was often very inexpensively constructed and, while standard in size, the bindings may not hold up well over time.

Bound volumes make up about 20% of the current holdings in the National Archives. The leather used for binding these volumes generally was not high quality and often on stationer's volumes and ledgers a combination of smooth vegetable tanned leather and suede was used, which inevitably deteriorated into dusty, "red rot." Some of the volumes are in fair to poor condition, with loose, detached or missing parts. Others while in good condition are vulnerable to damage due to their large to very large size, and great weight. In planning to move large numbers of damaged or potentially vulnerable bound volumes to the new National Archives facility at College Park, Maryland, the conservation staff was faced with the problem of safely relocating over 100,000 linear feet of bound volumes. One solution, specialized housing in hand crafted boxes, was rejected due to cost and the time required. Another solution was proposed. All volumes could be shrink wrapped, using technology familiar to anyone who has purchased pre-packaged meat or produce at the grocery store. A compromise was adopted. All volumes that would fit in standard boxes would be boxed. All other volumes would be shrink wrapped.

For almost a decade the National Archives had experimented with shrink wrapping volumes as a method of creating and maintaining clean stack areas. In the early 1980s, the Archives experimented with the process in the Central Plains and Southwest Regional Archives and in the National Archives Building in Washington, DC. Shrink wrapping of volumes was viewed as an expeditious method to: 1) reduce the level of dust in the stacks, 2) provide smooth clean surfaces for easy dusting, 3) prevent transfer of red rot leather dust onto everything in the vicinity of deteriorating leather bindings, 4) hold together loose or detached boards or spines, 5) hold volumes upright and aligned on the shelf, and 6) facilitate ease of handling when pulling them off the shelf and re-shelving them, thus reducing the chances of further damaging them.

At the time, no research had been conducted on the long term preservation implications of shrink wrapping and there were concerns that sealing acidic paper and leather in plastic films might result in accelerated chemical deterioration within the sealed package--a fear that a bound volume might "stew in its own juices." A preliminary study by the Library of Congress seemed to suggest that sealing volumes in plastic led to increased damage after accelerated aging. But the study was flawed by possible cross contamination between controls and samples aged together in the same oven and a choice of paper and pyroxylin binding that did not match the types of nineteenth century volumes predominating in the holdings of the National Archives.

At the end of the 1980s, the National Archives began a carefully designed accelerated ageing study of shrink wrapped volumes using acidic paper and discarded period bindings. Knowing that, regardless of the outcome of the new tests, shrink wrapping could be easily removed, the Office of National Archives chose to proceed with the use of shrink wrap as a temporary move preparation. The research and testing laboratory staff reviewed commercially available shrink wrap films and chose the film with the most stable chemical composition, DuPont Clysar EHC, a polyethylene/polypropylene copolymer which is FDA approved for wrapping food.

To prepare the volumes for shrink wrapping, the archival staff first identifies each volume. The identification includes the series title, inventory entry number, and the individual volume's sequence in the series (i.e. 1 of 6). The information is recorded on a long v-fold tab of archival paper which is slipped over the front board of each volume. The tabs are made of a heavy weight paper chosen for good durability that contains an alkaline reserve of calcium carbonate. While as yet untested, it is assumed that the chemical composition of the tabs may reduce the acid levels inside the wrap by absorbing possible acidic gases inside the package derived from the papers and/or binding materials. The identification process will assist the contract movers in re-shelving the material after it has been moved and greatly reduce the possibility of orphan volumes.

Before wrapping, archivists review the identification tabs for each volume, input that information into the automated master locator system designed for the move, and produce a form listing which volumes are ready for shrink wrapping. The conservation staff retrieve the volumes, brush off any dust, and tie those with loose or detached parts with white cotton twill tape. Flimsy volumes, volumes with missing boards, and those with sharp projecting parts receive stiffening layers of archival corrugated board tied with twill tape.

A loose-fitting sleeve of plastic film is created by inserting the volume between layers of folded film. The operator then closes an L-sealer which heat-seals the remaining open sides of the film. After sealing, the volume is placed on a belt that moves through the shrink wrap tunnel for several seconds exposure to oven temperature which softens the film and causes it to shrink into a close fitting wrap. While the temperature in the tunnel is fairly high, approximately 325 degrees Fahrenheit, the exposure time is limited and the heat scarcely penetrates inside the covers of the volume. A quick test using a temperature detector inside a paper-back book demonstrated this. The amount of shrinkage can be controlled by reducing or increasing the heat or the length of time the volume remains in the tunnel. It is important to use archival corrugated board to stiffen flimsy volumes or to protect volumes with missing boards as unsupported materials might suffer some distortion. Some archival volumes are very large and will not fit through the tunnel. In these instances, the film must be shrunk using a hand-held heat gun.

Some modifications of the shrink wrap technique have been made to allow shrink wrapping of non-bound records such as bundles of folded blueprint maps, stacks of blueprints and brownlines, and stacks of Census schedules tied together with flimsy cardboard wrappers. Archival corrugated board, cut and crimped into "C" or "L" shaped wrapper pieces, is placed around the material and tied with twill tape prior to shrink wrapping. Because no research has been conducted on the effect of the shrink wrap process on photographs, current policy is not to shrink wrap volumes containing photographs.

Volumes that are heavily used for research will not be shrink wrapped until just prior to relocation. Other volumes which may be needed for research are being shrink wrapped. These volumes can be unwrapped for researchers and then subsequently rewrapped. The unwrapping of volumes after the move will initially be limited to volumes containing color. This decision is the direct result of the controlled accelerated aging tests conducted by the Research and Testing staff in the Preservation Policy and Services Division. The initial results of these tests follow.

A STUDY OF SHRINK WRAPPED, SIMULATED BOUND VOLUMES
ABSTRACT

The National Archives and Records Administration undertook a study of the effects of shrink wrapping on paper in some simulated bound volumes. Four bindings removed from textblocks were cut in half perpendicular to the spine. In the control group paper was inserted in one half-binding, loosely tied, and aged for 8 weeks at 70°C and 65% relative humidity. Paper was inserted in the other half-binding, tied, then the package was shrink wrapped and aged for the same time under the same conditions. The properties of the sheets which were inserted in the binding halves and aged without shrink wrapping were compared with those of the sheets inserted in the matching halves and shrink wrapped before aging. This study tested the brightness, folding endurance, pH, and viscosity of these sheets. It was found that the brightness of the sheets in the bindings which were shrink wrapped decreased about 11 brightness units more than the sheets which were aged in the bindings without shrink wrapping.

INTRODUCTION

Shrink wrapping at NARA

The National Archives and Records Administration Regional Archives in Kansas City began shrink wrapping bound volumes in 1977 ¹. The record material chosen for wrapping was bound volumes "whose leather bindings were afflicted with red ...rot or whose canvas covers are in ribbons". Bindings which were no longer intact, or boards or spines which had become, or were in danger of becoming detached from the text block were also chosen for shrink wrapping. The packaging enabled these volumes to be stored as a unit on the shelves and cut down on the dust in the stacks. By 1984, the Kansas City Archives had wrapped 2,400 volumes and the Fort Worth Regional Archives had contracted for the shrink wrapping of 7,400 volumes. Thus, by April of 1984, two regional archives of the National Archives and Records Administration had turned to shrink wrapping as a means of securing bindings to text blocks and reducing dust in the stacks.

The film used in the Kansas City Regional Archives was DuPont Clysar EHC and the thickness was 150 gauge. The sealer and tunnel were contemporary Bestronic products; the sealer was hand operated. The tunnel was adjusted to 275oF to shrink the film. The residence time in the tunnel was on the order of 5 seconds.

Early Concerns and Testing at the Library of Congress It is apparent from Perry's article¹ that the question of potential long term damage resulting from sealing these materials in a shrink package had been raised. In fact, in 1980, the Library of Congress Research & Testing Office was commissioned by the National Archives and Records Administration to carry out research on this question. A memorandum on the subject "Effect of Shrinkwrapping with Plastic Film on the Aging of Books and/or Papers. Research Proposal No. 8, Report No. 22" summarizes the study². Two books of blank paper, different papers, but both acidic, were used for the study. The book cloth of these volumes was coated with pyroxylin, a cellulose nitrate coating commonly found on 20th century book cloth. The volumes were cut in half perpendicular to the spine, a few sheets removed to provide unaged control sheets, and half of each volume was shrink wrapped in Clysar EHC at the Kansas City Regional Archives. The shrink wrapped halves and the control halves were aged simultaneously in an oven at 90°C and 50% RH for 14 days. After conditioning at 23°C and 50% RH for one week, all the sheets were tested for pH, titratable acidity, brightness and fold endurance (modified to 0.5 kg.). The most dramatic difference between the unwrapped and wrapped sheets was the drop in folding endurance. However the significance of this difference cannot be assessed given the dependence of measured values for folding endurance on moisture content, since it is not known whether the individual sheets in each of the halves were equilibrated at 90°C and 50% RH before shrink wrapping the one half and aging both, or not. If the pages of each of the halves were not pre-equilibrated before the one half was shrink wrapped and both were placed in the oven and aged, the relative humidity in the oven would have been transferred to the pages of the two halves at a rather different rate because of the semi-permeable film surrounding the one half, and neither half, strictly speaking, would have been aged at 50% RH. Other measured properties showed less dramatic changes: brightness decreased, pH decreased, and acidity increased.

(Percentages in parentheses are the percentages of the properties retained.)

A note added to the data table indicated that the tests were run on identical pages from the wrapped and unwrapped halves of the books. However, the data presented shows an overall average for the property measurements, so any differences which may have existed between interior and exterior pages would have disappeared into the average.

In summary, the conclusion drawn by the Library of Congress researchers was that there was significant degradation of the aged shrink wrapped papers relative to those aged without being shrink wrapped, but that the protection afforded the materials might more than compensate for this increase.

While the conclusions drawn from the Library of Congress test appear to be supported by their data, there were some variables neglected which may have influenced the results. The books which were used for the study were both modern volumes with a pyroxylin-coated book cloth not typically found in the National Archives holdings. Pyroxylin is a cellulose nitrate having 11.5 to 12.3% nitrogen³ It has good dimensional stability, low water absorption, and toughness; its chief disadvantages are flammability and instability to heat or light⁴. The heat induced degradation results in the production of oxides of nitrogen³. These oxides in the presence of moisture give rise to acids which may contribute to the degradation of paper ⁵. While the Library recognized that this induced acidity may have contributed to increasing the rate of degradation, they stated that their "previous results with samples [of paper] sealed in glass had no pyroxylin present and still showed greatly increased degradation rates, so the effect of the pyroxylin cannot be considered as the major effect". Without knowing more about these earlier studies it is impossible to say whether their results obtained on papers sealed in glass have any bearing on these results from modern volumes sealed in semi-permeable film. Furthermore, without permeability data for oxides of nitrogen through Clysar EHC one can't be certain that the trapping of pyroxylin degradation products was not a major factor in the loss of properties seen in the Library of Congress study.

Without greater knowledge of the binding materials in the NARA holdings, it is difficult to compare and contrast the effects expected from the shrink wrapping of volumes in the NARA holdings with those obtained from shrink wrapping the volumes studied by the Library of Congress. Not only did they study only one type of binding, but in all of these studies, we have examined blank test paper. Thus we can only speculate as to what differences might be seen when shrink wrapping actual record materials. Since the use of pyroxylin for treating fabrics has been dated to about 1910, we can't discount the possibility that the Archives does have some pyroxylin-coated or impregnated book cloth⁶.

In all likelihood, because the material in the Archives holdings is on the average much older than the modern volumes chosen for the Library of Congress study, degradation of Archives materials is probably well underway before shrink wrapping, and thus we would see much less degradation due to pyroxylin degradation products. This probable prior degradation, coupled with the possibility that the degradation is exponential (in which case a majority of the our holdings have reached a plateau, or a state in which the rate of degradation has slowed appreciably) would suggest that degradation of new pyroxylin-coated bindings would proceed more rapidly and result in more degradation products than pyroxylin-coated bindings in Archives holdings. Moreover, the binding material in our holdings, which does not contain cellulose nitrate, will likely degrade by different mechanisms (about which little is known).

There is a large body of evidence which suggests that the initial product in the degradation of cellulose nitrate, the NO2 free radical, participates in the second step of the degradation reaction ⁷. Thus the degradation of cellulose nitrate could be viewed as autocatalytic, and more likely to proceed faster than many other degradations.

Among the questions as to how the study was conducted are ones concerning the treatment of the volumes prior to and during aging. When material is subjected to an increased temperature and relative humidity for the purposes of artificially aging, every attempt is made to insure that the material experiences that temperature and relative humidity evenly throughout. For a bound volume this would require pre-equilibration of the bound volume with the pages propped open until each achieved the same moisture content. Then the volume would have to be quickly sealed. The exact same equilibration would have to be carried out with the volume which was not destined to be sealed despite the fact that the paper in it seems to be exposed directly to oven conditions. A body of data exists which suggests that center sheets in stacks of paper or the center of paper rolls achieve moisture equilibrium very slowly⁸. The report does not suggest that any pre-equilibration took place, in which case sheets from different parts of the volumes were subjected to different relative humidities, and the sheets in the shrink wrapped volumes were probably not aged at the same relative humidity as those in the control. Also the Library researchers reported that the unwrapped and wrapped halves were aged at the same time in the oven. This is not the usual practice. Controls are generally not aged in the same oven at the same time as the experimental sheets because of the possibility that the experimental treatment, in this case, the film, may have an adverse effect on the controls.

There are also a number of questions regarding the treatment of the data. The research reported here, as well as further work at the Library of Congress, indicates that properties of pages may change in relation to their position within the volume with aging ⁸. Therefore the properties of a number of sheets from different positions in a book should not be averaged without testing whether those properties show less variation from sheet to sheet than within the sheet. It does not appear, despite the fact that the Library of Congress researchers said that "comparisons were made on identical pages from each book half", that these comparisons were reported. Instead the numbers reported appear to be averages of the properties tested determined using an indeterminate number of pages for each half of each book. Had they compared the same pages from each half as they indicated they had, an average difference between the wrapped and not wrapped halves would have been given, or individual differences between groups (or even between pages), would appear if it was statistically unacceptable to average all the individual differences. The folding endurance values reported are not identified as to whether the samples were cut with the paper grain or against it. The TAPPI method for determining folding endurance requires that 10 representative samples in both the cross and machine direction of the paper be cut from each test unit⁹. Because 8 1/2 x 11" sheets cannot be cut to give the requisite number of machine direction and cross direction samples, and the sheets for the Library study were obtained by cutting two ordinary size volumes in half, it is unlikely that 10 machine direction and cross direction samples were obtained from each sheet. This is not unusual, and when it occurs as many samples as possible are usually taken from each sheet in each direction and a number of sheets are used to make up the test unit. However, when this occurs a check should be made to insure that the folding endurance does not vary greatly from sheet to sheet or the concept of test unit is invalid. In the case of bound volumes which have been aged, it appears that, at least in the vicinity of the covers, the sheets vary considerably; but we have also tested unaged papers which showed a greater variability in fold from sheet to sheet than within the sheet. Also, because the number of measurements making up the averages are not reported, the significance of the data cannot be assessed or compared with other data.

There are very few reports on research undertaken to determine changes in paper properties on aging in sealed containers (10,11,12,13). What exists tends to support the Library of Congress statement that sealing books or acidic papers in relatively airtight containers hastens their degradation because gaseous degradation products are sealed into their microenvironment. What should be noted about all of these early studies is that they examine individual sheets of paper sealed in either glass tubes or polyester envelopes, materials which have very little, if any, gas permeability. While the permeability of Clysar EHC to paper and book degradation products is not known, the oxygen permeability of 100 mil Clysar EHC is 325 cc/100 in2 24 hrs.atm./mil, while that of 100 mil Mylar is 6 cc/100 in2 24 hrs.atm./mil at 25°C (14,15), which leads to the conclusion that Clysar is much more permeable than polyester, and certainly more permeable than glass.

EXPERIMENTAL

Selection of bindings

The selection of samples for this study was a bit unusual. The bindings were salvaged from a disbinding project being carried out on the Papers of the Continental Congress in 1984. Although they are old and resemble a number of the bindings in the Archives holdings, they are not contemporary with the Papers of the Continental Congress and do not necessarily typify the bulk of our bound holdings. At some point before these materials were acquired by NARA from the Library of Congress, the records were hinged to support pages and bound. This approach to storing letters, etc., was common in the last century and probably continued into the early 20th century. Since the Papers of the Continental Congress have been in the custody of the National Archives they were deacidified, while still bound, by brushing methyl magnesium carbonate in 1,1,1 trichloroethane on the individual pages. When the bindings were chosen for this study little was known of their history. It was noted that there was a white haze on the inside of the covers which was distributed unevenly around some of the inside edges. The knowledge that the Papers of the Continental Congress had been deacidified while still bound explains this haze. The presence of alkaline buffer in the boards may have an effect similar to that achieved by using book label tabs with alkaline reserve in the ongoing NARA shrink wrapping of bound volumes, i.e. to mitigate the effect of any volatile acid degradation products.

The size, similarity, and the ease of attaining these bindings made them candidates for this study. Because their average dimensions were 14 1/4" by 18 1/8", each could be cut in half perpendicular to the spine and still enclose 8 1/2" by 11" pages. Three of the four volumes were half-bound in leather. The fourth cover was a grained book cloth. A consolidant had been used on the leather of the spines and some of the corners. Each of the bindings contained an made endsheet constructed of a marbled paper adhered to a plain sheet. The fly leaves remain attached to the spine by a linen hinge on the inside which was covered by the book cloth on the marbled side to resemble the inside edges of the cover. The insides of the covers retain the marbled paper paste-downs. The remaining volume (volume 2) was a later re-casing, using new boards, new book cloth, the old end sheets, with what appears to be the original linen hinge, covered on the outside with the new book cloth, and with the original spine reattached to the exterior of the new one. The stacks of test papers were placed within the boards inside the endsheets.

Choice of paper stock

The paper for the volumes was chosen with greater specificity. It was important to have a strong, alum-rosin sized, fully bleached paper. The alum-rosin sizing was essential because much of the paper in the Archives holdings dates from the time period when chemical wood pulp was alum-rosin sized. It was optimal to have a strong sheet so that properties such as folding endurance, which have inherently large standard deviations, would be able to decrease significantly without being essentially zero. An envelope paper meeting these criteria was supplied by Westvaco Paper Company from their Wycliffe, Kentucky mill. The samples used for the various parts of the experiment were chosen using a random number table.

Shrink wrap film

The film used for shrink wrapping was 75 gauge (0.75 mil) Clysar EHC. It is a polyethylene polypropylene copolymer containing no plasticizers¹⁴. The laboratory had investigated some of the other films available before beginning the study and judged it to be among the most stable. The Beseler "L" heat sealer and the Bestronic shrink tunnel were installed and adjusted just before the samples were shrink wrapped.

Oven aging conditions

Although there are three written official standards for artificial aging of paper1, the conditions chosen for this study did not match any of them. The conditions chosen for this study were 70°C, 65% relative humidity. This choice was made based on work carried out for the National Archives and Records Service by the National Bureau of Standards¹⁷. This study examined the degradation of a number of paper properties as a function of temperature and relative humidity. When either the value, or the log of the value, obtained at each of the temperatures and relative humidities was plotted against aging time a straight line was obtained. The slopes of these lines were calculated and called the degradation rates (for the various properties). When these degradation rates were plotted against relative humidity, the lines for all these rates passed through zero at zero percentage relative humidity; however, the degradation rates of the various properties obtained at 90°C and the degradation rates for some of the properties measured at 80°C were exceptions. The authors suggested one explanation for this might be "that the mechanism of degradation changes appreciably at some temperature above 70°C." While the authors' argument that the degradation rates at various temperatures should fall to zero when extrapolated to zero% relative humidity seem to contradict results obtained using dry oven aging, the possibility that the mechanism might well change above 70°C caused us to choose 70°C as the temperature for our aging of shrink wrapped books. Also, since the study of aging shrink wrapped "bound volumes" was viewed as self contained, i.e. the results would be compared with the volumes which were aged without wrapping, it did not seem necessary to have the data or conditions compare with those of previous studies.

Determination of testing conditions for the study

In order to determine the appropriate aging time for the paper in the volumes 87 sheets were hung in the oven and aged at 70°C and 65% relative humidity. An equal number of sheets were removed from various parts of the oven at 1, 2, 4, and 8 weeks and tested together with 14 unaged controls for brightness, pH, folding endurance, moisture content (to arrive at a dry weight needed for viscosity), and viscosity. Testing followed the TAPPI methods with the following exceptions: brightness was tested according to TAPPI T 452 (except that the sheet was not cut into tabs), six brightness readings were taken from random places on the sheet backed only with the 1 kg. weight; pH was measured by the cold extract method following TAPPI 509; the folding endurance was measured with an M.I.T. fold endurance tester following TAPPI 511, with a 0.5 kg. weight used to set the scale instead of the specified 1 kg.; TAPPI procedure T 412 was followed to determine the moisture content of the samples, and results used to correct the weights used in the viscosity measurements (done following TAPPI procedure T 230) to dry weight.

These tests were chosen primarily because they would afford a comparison to the Library of Congress results from the original evaluation of the effect of shrink wrap, and judicious use of a sheet could produce enough sample for all the tests with one minor modification, only 9 machine direction and 9 cross direction folding endurance samples could be obtained from one sheet: the method specifies 10. An additional test for viscosity, not part of the Library of Congress evaluation, was added. This test was chosen to indicate, approximately, changes in the length of the cellulose chains upon aging. As cellulose ages the degree of polymerization decreases; accompanied by a decrease in viscosity of a cellulose solution, or an increase in the speed with which the cellulose solution will flow between the two marks on the viscometer. The viscosity of a polymeric solution is, indirectly, a measure of the length of its chains, or its degree of polymerization; the longer the chains, the higher the degree of polymerization, and the slower the solution moves in a viscometer. The TAPPI measurement of viscosity correlates with the more precise work done by H. Burgess at the Canadian Conservation Institute using size exclusion chromatography to determine degree of polymerization and to follow its decrease upon aging as a direct measure of cellulose degradation ¹⁸.

The results at 1, 2, 4, and 8 weeks were compared with the results on the 14 unaged controls to determine the aging period required to show a significant difference. The conclusions reached were: that brightness changed significantly with each aging period, so that any of the time periods would produce a significant brightness change;folding endurance did not show a statistically significant difference until 8 weeks of aging; pH did not show significant differences at any of the aging times, which was not unlike what the Library of Congress researchers saw when measuring the pH of the Mead bond paper; and the viscosity changed significantly with each of the aging periods. Thus the volumes would have to be aged eight weeks to see significant changes in most of the measured properties if they behaved like the single sheets. Work cited previously⁸ indicated that the changes seen in the papers within the volume would be greater than those seen in the freely hanging sheets. Thus while it was expected that pH would change significantly in the bound volumes, it was also possible that it might not.

The previous work also suggested that one might see a difference in properties of the sheets based on their position within the bound volume, that is, that there might be a trend in the value of the properties through the volume. If this were the case then it would be invalid to average properties from sheets throughout the volume to arrive at an average property for that volume. In the extreme it might be true that the only valid comparisons which could be made would be page to page comparisons, i.e., only page one of the control half could be compared with page one of the corresponding half which was shrink wrapped. If this were the case, in order to see a trend, it would be necessary to test a number of volumes.

Preparation of volumes for aging

With this in mind seven volumes of 150 pages each were made. The volumes were made by hanging 150 sheets together with the binding in the oven at 70°C and 65% relative humidity for an hour, then while still in the oven, putting the sheets in the binding, and tying with high quality stainless steel wire. The volume was placed in the middle of a long piece of wire. The wire was wrapped once around the width of the volume, twisted around itself in the center, wrapped once around the height, twisted together by hand and cut with scissors. The tying was tight enough to keep the sheets in the binding when the bindings were placed on their edges in the oven, but probably did not apply much pressure to the pages, since the whole operation was carried out in the aging oven, by hand. Then the volume was moved to another oven at 70°C and 65% relative humidity. The procedure was repeated to create the six additional volumes.

Oven failure and loss of samples

During this time it was noted that the oven in which the volumes were being pre-equilibrated was not functioning properly, but because it was working while the pre-equilibration and assembling of the volumes was going on this was not regarded as consequential; it just meant that all the volumes would have to be aged in the oven to which they were moved after being made. However, this was to prove fatal, since that oven failed partway through the eight weeks (it did not have a temperature, relative humidity chart recorder, thus the exact time of failure couldn't be known), and because there was no working back-up oven to which the volumes could be moved, the samples were essentially lost.

Preparation of samples used in study

At this point only enough sample remained to make four volumes of 50 sheets each which would not be wrapped and four volumes of 50 sheets each which would be wrapped. The four volumes designated as the controls were pre-equilibrated and assembled at the same time in the oven in which they were aged (by this time the earlier problem with this oven had been solved). Assembly included tying with stainless steel wire to keep the pages in the bindings when the volumes were propped in the oven. At the end of eight weeks these volumes were removed, placed in the TAPPI room, disassembled, and tested. By the time the volumes that were to be shrink wrapped were assembled, the National Archives and Records Administration had a working shrink wrap sealer and tunnel in Washington, D.C.. Thus after the volumes were quickly assembled and tied, they were placed in specially made polyester bags, sealed, carried to the machine, unbagged, sealed in Clysar EHC, the film was heat shrunk, the volumes were re-bagged and returned to the oven to age (with the polyester bags removed). The time frame for shrink-wrapping the volumes and returning the to the oven was kept as short as possible (this entire process took about a half an hour) so that they were not in an atmosphere different than that at which they were to be aged any longer than necessary. The volumes were supported in the oven on the bottom edge and the top two corners by placing them against the bottom and sides of a stainless steel screen cage placed inside the oven. In the experiments carried out prior to the actual shrink wrap study, sheets were removed from various parts of the oven, and their properties tested, not only to determine how long the aging should be carried out but also to be sure that the volumes placed in various parts of the oven would age the same.

RESULTS

The brightness, pH, folding endurance, and viscosity measurements tables are available upon request from Susan Lee-Bechtold, Supervisory Chemist, Document Conservation Branch, Preservation Policy and Services Division, National Archives and Records Administration at College Park, 8601 Adelphi Road, College Park, Maryland, 20740. The measurements are listed by the page number under the respective volume from which they came. Properties were measured on the first 6 pages, the last 6 pages and the middle 12 pages in each volume. This was done so that if it were found that the volume as a whole showed a trend throughout with respect to the properties measured, it might be possible to group the outside pages and compare them with the inner pages. The results are also plotted on the graphs which follow. The graphs, if not the numbers themselves, make it clear that the only valid comparisons for most of the properties are between the pages in the same position in the control (or not wrapped) volume and in the companion wrapped volume.

DISCUSSION

Brightness

The shrink wrapping caused a statistically significant decrease in brightness of about 11 brightness units or a decrease of about 15%. In a paper as white as the test paper was, neither the control nor the wrapped sheet is as white as the unaged paper, but the shrink-wrapped sheet appears to have darkened slightly more than the control. There is also an effect of position on brightness, both in the controls and the shrink wrapped papers, and on the differences between control and shrink wrapped paper. An examination of the plot of brightness versus position for the control and shrink wrapped sheets of all four volumes shows that the brightness increases from the beginning sheets to the middle sheets, then decreases from the middle to the end sheets. The magnitude of this effect seems to be the same for both groups. If this is the case, when the differences between the brightness of the controls and the respective shrink wrapped pages are plotted the effect of position should disappear. Plots, which can be obtained from the above source, labeled Figures 1,2,3,4, and 5 , show that, indeed, the differences do disappear for the papers in the middle and end positions. In the beginning positions the scatter due to the outliers in the controls does not allow us to draw definite conclusions about the effect of position on the differences between the control and shrink wrapped sheets. While it would be convenient, removal of the outliers cannot be justified for experimental or statistical reasons¹⁹.

Figure 6 reveals a further important feature. The brightness differences between the control and shrink wrapped sheets also show an effect that seems to be related to the volume from which the sheets came. Thus it appears that the differences are smallest for volume 1, somewhat larger for volume 2 and 3 and largest for volume 4. This effect is most pronounced for the middle and end sheets. Again any effect of this sort is obscured in the beginning sheets by the outliers in the control sheets. The differences seen cannot be related to any visible differences in the volumes, since from visual inspection the bindings for volumes 1, 3, and 4 appear to be the same, while volume 2 is one which looks different, with new boards and a different book cloth.

Again, the conclusions to be drawn about brightness are:

1. The shrink wrapped sheets are about 11 units of brightness lower than the corresponding control sheets.
2. Both in the controls and in the shrink wrapped sheets there is a systematic difference between the 4 volumes.
3. Position has a small but definite effect on brightness, both in the controls and in the shrink wrapped sheets. The brightness increases from the beginning to the middle, and decreases from the middle to the end by about 2-3 units on all 4 volumes.

Thus the only valid comparisons to be made, at least with respect to brightness, are the comparisons between the individual control and shrink wrapped sheets in the same position within each volume. Thus the significant numbers are the averages of the differences calculated from the respective brightness differences between these individual sheets.

Folding Endurance

The folding endurance results do not show any discernable differences between the samples which were wrapped and those which were not. The results, displayed to accentuate differences between the same respective pages in the wrapped and control volumes, are available as tables and have been plotted as figures. The differences have also been plotted. The tables and graphs are available from Susan Lee-Bechtold, Supervisory Chemist, Document Conservation Branch, Preservation Policy and Services Division, National Archives and Records Administration, Washington, DC 20408. It is apparent that the trend seen in brightness doesn't appear in fold endurance. Only the first 6 pages of volume 1 measured in both the machine and cross direction, the middle 12 pages of volume 1 measured in the machine direction, the first 6 pages of volume 2 measured in the cross direction and the last 6 pages of volume 4 measured in both directions show average differences which are unquestionably greater than zero. A number of the sets have instances when the folding endurance of the shrink wrapped sample was greater than that of the control, together with measurements showing the opposite. Another table, Table 4, synopsizes the machine and cross direction folding endurance for the controls and wrapped sheets divided into top, middle, and bottom groups. Given the variation within the sheet, treating the data this way is statistically valid. However, even examining the data this way leads to the same conclusion. While there are trends within volumes, the data does not support the hypothesis that shrink wrapping has either a positive or negative effect on the folding endurance measured in this study. In fact, given the variability within the sheet, within the groups, and within the volumes, it would also be valid to average all the machine direction folding endurance values, and all the cross direction folding endurance values. However it does not yield any additional information.

pH Values

Tables 5 and 6 express the pH measurements in a number of different ways. Again, the data does not permit differentiation between the pH of the wrapped and the control sheets. There are slight differences from volume to volume, but these do not appear consistent. Available are figures 17-20 which present the individual pH measurements and a graph of the differences is available as Figures 21-24. The tables and graphs are available from Susan Lee-Bechtold, Supervisory Chemist, Document Conservation Branch, Preservation Policy and Services Division, National Archives and Records Administration, Washington, DC 20408.

Viscosity

The TAPPI method for determining viscosity requires two samples of exactly 0.2500 g moisture free pulp for each measurement. Because of the amount of time required to arrive at the requisite number of samples of exactly 0.2500 g moisture free pulp it was suggested that we use two samples very close in weight when dry to 0.2500 g, but one larger than that value and one smaller, and plot the calculated viscosities for these samples in order to arrive at the viscosity for a 0.2500 g sample. We ascertained, on the original sheets, that within the fairly narrow range of 0.23 to 0.27 g the viscosity of the paper was linear. Then we tested the method on the sheets hung in the oven and aged for 2, 4, and 8 weeks. The viscosity of the original unaged sheets was 10.50 ± 0.69 mPa sec, at two weeks it was 7.08 ± 0.096 mPa sec, at 4 weeks 5.90 ± 0.095 mPa sec, and at 8 weeks 4.88 ± 0.061 mPa sec. Thus despite the simplification being made (that all the viscosities could be calculated using the density of the 0.2500 g pulp solution, rather than the unknown densities which would have been slightly different for each sample) the results were quite consistent. The results are reported in a table, Table 7, which lists the average times of efflux, the weights, the calculated viscosities, and the viscosities determined from each graph for a 0.2500 g dry weight sample of that sheet. (Also available are Figures 25 and 26 showing the viscosities for the control and wrapped sheets, respectively.) The tables and figures are available from Susan Lee-Bechtold, Supervisory Chemist, Document Conservation Branch, Preservation Policy and Services Division, National Archives and Records Administration, Washington, DC 20408.

It was expected that there would be a significant decrease in viscosity in the wrapped sheets, thus measurements on the shrink- wrapped sheets were begun with a larger number viscometer (which actually has a smaller capillary). When the times of efflux came out to be so large (i.e., 310 seconds, etc.), and the calculated viscosities about the same or only slightly lower than the control sheets, we changed back to the viscometer used previously. While, in some groups, in some volumes, there appear to be non-zero differences between the control and wrapped sheets, these differences are not consistently either positive or negative. For example, in Volume 1 the difference between the control and wrapped top sheets was 0.51 ± 0.20 but the bottom sheets showed an average difference of - 0.38 ± 0.17. The remaining differences in the other volumes and volume 1 are so close to zero as to make any conclusions drawn about differences between the wrapped and control sheets' viscosities questionable.

A close inspection of the graphs of viscosity for the various volumes might permit one to draw lines through the values for the initial five and final five sheets. An inspection of these lines might lead one to conclude that the viscosity of the wrapped sheets was higher than that of the controls. Because of the scatter in the initial and final values this doesn't seem to be supported statistically.

RECOMMENDATIONS FOR FURTHER RESEARCH

While the study demonstrated that the shrink-wrapped sheets darkened significantly relative to the control sheets, it is difficult to interpret this change in chemical terms. The literature seems to suggest that you can experience a change in brightness, resulting from an increase or change in chromophoric groups in the cellulose molecule without seeing a change in the viscosity (20,21,22). Thus it is possible that the aging inside the shrink-wrapped volumes produced more carbonyl groups, decreased the alpha-cellulose content, or caused some other change we did not detect, without immediate cleavage of any cellulose chains. This does not imply that an increase in carbonyl groups in the cellulose molecule would not ultimately lead to a decrease in the length of the chains. Since the carbonyl group is generally more reactive than the alcohol group from which it was made, this replacement should increase the cellulose reactivity and ultimately lead to chain degradation. For this reason it seems wise to undertake the measurement of alpha cellulose and/or carbonyl content on the remaining fragments of the control and test sheets first.

Then, there are at least two further issues: first, what is the effect of aging at longer intervals, or different conditions;and second, what effect does the shrink wrapping process itself have. To answer the question as to what happens when shrink wrapped papers are aged for longer times, we should continue aging the remaining sheets in the control and shrink-wrapped volumes and measuring their brightness both before and after aging. Because the results so far suggest that sheets from the shrink wrapped half of volume 1 match can only be compared with sheets from the half of volume 1 that wasn't wrapped, and that these results don't necessarily predict how the sheets in volumes 2, 3, or 4 will behave, aging longer to see if the degradation of brightness continues would need to be done on both halves of all four volumes. Because there is the question as to how the actual wrapping process influences the differences we saw, it would be desirable then to age all four controls, with the remaining 25 sheets in each, and all four shrink wrapped halves, with their 25 respective sheets, separately, for one longer period, rather than removing the volumes, unwrapping, reading brightness, re- wrapping, aging longer, unwrapping, checking brightness and then re-wrapping for further aging. Choosing only one longer time period will not enable us to extrapolate brightness as a function of time of aging, but it will tell us whether the brightness difference will increase, decrease, or remain constant as aging progresses. This can be done without destroying any of the remaining test papers.

The other influence we should examine is the effect of the shrink wrapping operation itself. Since the effect of re-wrapping and shrinking after each measurement cannot be ascertained from the study described above, it is important to carry out an additional study. While it would be desirable to use the same paper, if we had more, it is possible to study the effect of the shrink wrapping process independently. For this study the paper, or papers, would not need to be chosen so carefully, as we would merely be testing brightness changes. We would expect to create volumes as we did previously, using an alum-rosin sized, acid paper, measure the brightness of the papers, wrap the one half, unwrap, test brightness on both the wrapped and control, re-wrap, and go through this process at least 5 or 6 times. At the completion of this portion of the study we would age the individual sheets, in two groups, the controls together, and those from the shrink wrapped volumes as a separate group, to be certain the wrapping had not caused some change that we did not detect immediately. Given the small differences seen in the study just completed, it may be impossible to ascertain the contribution of the shrink-wrapping process itself to the degradation of the paper within the volumes. But we should try. If there were no significant changes after shrink-wrapping a number of times it would be safe to assume that the changes in the aged shrink wrapped volumes are related to the microenvironment, rather than to the act of shrink wrapping.

A number of other suggestions for further research have been received. To pursue any, or all of these, we must find a source of test paper (which we have been trying to do since before the test volumes emerged from the oven). Assuming that we find suitable paper, it has been suggested that we test the deterioration of books with pyroxylin coated binding as well as those with red rot, shrink-wrapped both with and without alkaline tabs. We might also naturally age some test volumes, testing them at various times. It has also been suggested that the experiment carried out in this study be done using different aging temperatures, and for different lengths of time. Moreover a suggestion was received that we try to simulate the environment that bound volumes experience packed tightly onto shelves, because the degradation in this case might not be greatly different than the shrink wrapped volumes in the study. The only problem with this suggestion is that we also pack our shrink- wrapped volumes just as tightly on the shelf as ones which have not been wrapped, and it's difficult to achieve this close packed condition, with little air circulation, in a humid aging oven. However if we were able to achieve this tight packing for both the controls and the shrink-wrapped volumes, it would be valuable to examine whether the differences between them were as great. Since one of the alternatives to shrink-wrapping is packing in a box, it would also be desirable to compare how volumes age in boxes. Given the complexity of the results obtained, it is important that one or more of these suggestions for further research be pursued.

1. ASTM D776, TAPPI T 453-pm85, and ISO 5630/1 are all designations for a dry oven aging method carried out at 105 ± 2°C. ASTM 4714-87, TAPPI T 544-pm-85, and ISO 5630/2 are all designations for a humid aging method in which the temperature is 90°C and the relative humidity is 25%. The third is and international standard ISO 5630/3 which specifies a temperature of 80°C and a relative humidity of 65%. This standard was designed to simulate aging in tropical climates. Despite the fact that the conditions in the U.S. humid aging standard are 90°C and 25% relative humidity, they are rarely used. It is laborious to achieve the 25% relative humidity as most humid aging ovens do not operate at that low a relative humidity. Thus,it has evolved that most humid aging in the U.S. is done at 90°C and 50% relative humidity. It is expected that TAPPI will revise its humid method soon to reflect this practice¹⁶.

References

1. Perry, A. 1984. Packaging the Problems in Kansas City, The Abbey Newsletter, 8(2), 25.

2. Kelly, G.B. 1980. Effect of Shrink-wrapping with Plastic Film on the Aging of Books and/or Papers. Unpublished report. Washington, D.C.: Library of Congress.

3. Reilly, J. A. 1991. Celluloid Objects: Their Chemistry and Preservation. Journal of the American Institute of Conservation 30(2), 145-162.

4. Rich, R. P. 1980. Modern Plastics Encyclopedia, 57(10A), 25.

5. Williams II, E. L., and D. Grosjean. 1990. Exposure of Deacidified Paper to Ambient Levels of S02 and N02, Getty Scientific Program Report, July 1990.

6. Roberts, M.T. and D. Etherington. 1982. Bookbinding and the Conservation of Books. A Dictionary of Descriptive Terminology. Library of Congress, Washington, D.C., 32.

7. Shashoua, Y., S. M. Bradley, and V. D. Daniels . 1992. Degradation of Cellulose Nitrate Adhesive, Studies in Conservation, 37, 113-119.

8. Nordman,L. 1974. Influence of Environment on Paper and Board Properties, Papper. Och. Tra., 1974(3), 123. And Shahani, C. J. 1991. Presented at the IFLA and ICA sponsored International Seminar on Research in Preservation and Conservation, Harriman, New York, May 1991.

9. TAPPI. 1992. Folding endurance of paper (MIT tester), T 511 om-88. In TAPPI Test Methods. Atlanta:TAPPI Press.

10. Santucci, L., and M. G. Zappala Plossi. 1973. Invecchiamento della carta in tubo chiuso. In Problemi di Conservazione, Parte II, Ed. Compositori, Bologna. 501-512.

11. Browning, B. L,. and W. A. Wink. 1966. Investigation of the Durability and Permanence of Writing Papers, Reports 2 and 3. A Progress Report to The Cotton Fiber Paper Group Writing Paper Division American Paper Institute by the Institute of Paper Chemistry, Appleton, Wisconsin.

12. Hengemihle, F. H., N. Lindsey and C. J. Shahani, C. J. 1989. AIC abstracts, 17th Annual Meeting, American Institute for Conservation, Cincinnati, Ohio. 59.

13. Unpublished data. 1980. National Archives and Records Service, Washington, D.C.

14. Du Pont Company. Clysar. Safety in Handling and Use. E- 66324.

15. E. I. Du Pont De Nemours Co. (Inc.). Mylar. Technical Information. Chemical Properties.

16. Wilson, W. K. 1992. Personal communication.

17. Graminski, E. L., E. J. Parks, and E. E. Toth. 1978. The Effects of Temperature and Moisture on the Accelerated Aging of Paper, NBSIR 78-1443.

18. Burgess, H. D. 1982. Science and Technology in the Service of Conservation. Preprints of the Contributions to the Washington Congress (IIC). 85-88.

19. Natrella, M. G. 1963. Experimental Statistics, National Bureau of Standards Handbook 91. Washington, D.C.: Department of Commerce. Chpt. 17.

20. Rochas, P. 1957. The Chemistry of Cellulose Degradation. Bull. Inst. Textile. de. Fr., 65, 15-40.

21. Stuebchen-Kirchner, H. 1962. The Yellowing of Cellulose. Osterr. Chemiker-Zeitung, 63(10), 319-330.

22. Spinner, I.H. 1962. Brightness Reversion. TAPPI, 45(6), 495-514.

NOTE: Data and graphs are available from Susan Lee-Bechtold, Supervisory Chemist, Document Conservation Branch, Preservation Policy and Services Division, National Archives and Records Administration, Washington, DC 20408.