WWII Japanese Flag Post Treatment Report

POST-TREATMENT REPORT

Goals of Treatment:

To stabilize the degraded silk and prevent further deterioration
To improve visual appearance and prepare for display

Overview of Treatment:

Vacuuming
Poultice treatment
Wet cleaning with Orvus foam on a suction table
Blocking and drying
Adhesive backing, possibly
Mounting

Vacuuming:

The flag was vacuumed with a variable suction MuseumVac™ on low suction, through a screen, using microattachments to remove surface dirt and particulate buildup.

Poultice Treatment:

Due to the physical dangers to the nervous system through inhalation of m-xylene, the treatment was required to be conducted under the fume hood instead of on the suction table in the wash room as was originally planned. Upon placing the flag in the fume hood, it was decided that only a spot treatment of the staining would be attempted rather than a poultice treatment of the entire flag as had been suggested in the treatment proposal. The wish to limit the conservator’s exposure to the m-xylene, coupled with the fume hood not being large enough to unroll the flag completely, made a spot treatment the best option. The air flow in the fume hood was strong enough that it was causing an unacceptable amount of movement; this could have caused further damage to the flag despite the flag being rolled in a scroll-like manner on two tubes to prevent this type of movement. The hood was raised enough to reduce the amount of air flow moving the flag, but could not be raised entirely due to the dangerous fumes emitted by m-xylene.

A poultice of m-xylene was applied to the flag using cotton pads of a 6cm diameter and 4mm thickness and a dropper. The smaller stains received 3 drops and the larger stains 5 drops. A layer of blotter paper was placed underneath the flag to absorb extra solvent and catch the particulates which had been flushed out of the fabric; this helps prevent tide lines caused by redistribution of the particulates. A glass plate was placed over the cotton pad to prevent the solvent from evaporating too quickly and to facilitate better contact with the flag. Though testing showed no dye bleed when in place for 20 minutes, the poultices were only left for 10 minutes as this was sufficient to remove any visible particulates in most areas.

In the most soiled and stained areas the flag was lifted slightly (1cm) and m-xylene was applied with a dropper directly to the flag. The gravitational force pulled the remaining black-brown buildup down through the weave structure and out onto the blotter paper underneath. The flag was then placed back on the blotter and a clean cotton pad was applied to help pull any remaining particulates from the fabric.

This method was used to remove the significant buildup from the center of the red sun. While the dye had not moved during original testing or during the original 10-minute poultice treatment, when the clean cotton pads and blotter were checked a significant amount of red dye could be seen on the blotter (fig.1). The cotton was immediately removed and the blotter was changed. No further attempt was made with this treatment and the white sections of the flag had already been completed. Due to the fact that there was no change to the color of the red sun, it is likely that the m-xylene was simply reactivating and flushing out the excess dye from the printing process. However, it was not worth the risk of the dye bleeding into the rest of the flag to continue. The original 20-minute test was performed on such a small area that there might not have been enough solvent or excess dye to create the same conditions.

Fig.1 – Blotter paper from under the flag during m-xylene poultice

Wet Cleaning:

After the m-xylene had ample time to evaporate in the fume hood, the flag was moved to the suction table for wet cleaning. Mylar™ sheets had been cut to allow movable 7cm sections of the flag to be exposed. The flag was rolled scroll-style on two blue board tubes covered in soft Tyvek™ to prevent any water or Orvus from reaching the blue board core. This also facilitated exposing only small sections at any given time. The flag was worked from the right to the left, keeping the paper edge supports protected from any possible splash or spray until that section was exposed (fig.2). A sheet of clean Reemay™ interleaving was used as the flag was rolled onto the new tube to prevent any sticking or friction which might cause further damage.

Fig.2 – During wet cleaning with Orvus foam on suction table

The flag was cleaned in barely overlapping 7cm sections by first dampening the area with a lab bottle filled with distilled, pH balanced water. A 5% w/v Orvus in water solution had been prepared in a foaming hand soap bottle to allow for a consistent, dense foam, as it was the bubbles bursting that would provide much of the mechanical action. The fabric was too fragile to withstand any stronger mechanical action without causing further damage. Once the area was wetted, a layer of Orvus foam was applied, spread gently with a finger, and allowed to sit for 5 minutes. At this point the suction table was turned on to its lowest setting to pull the remaining Orvus foam through the weave, providing more mechanical action. This step was then repeated. While the suction was still on, the area was flushed 3 times with the water in the lab bottle and the area was totally covered for a moment to maximize suction and flush out any remaining Orvus or water.

This process was repeated by rolling the still damp flag onto the receiving tube until the opposite, or left, side of the flag was reached. When the paper edges were visible, they were wrapped in Mylar™ and the areas being addressed were made smaller, approximately 2-3cm, near the supports to ensure no water came into contact with the paper. Once this entire area was completed, it was rinsed again with water in the lab bottle. Afterwards, a piece of blotter was placed across the entire length of the flag and covered with plastic to facilitate drying and removal of any remaining particles or dirt. When this section was dry, the flag began to be rolled back onto its original tube and each section was rinsed a final time before it was dried and rolled back onto the original tube (fig.3).

Fig.3 – Drying with blotter paper on suction table

Due to the very fragile nature of the flag when it was wet and the inability to see through the blotter, there were some areas which dried with small creases. To remedy this, the areas were gently moved back into place and small pieces of blotter were placed over the creases, dampened with 2-3 drops of water, and set under a glass weight for a few minutes. This effectively removed the creases caused during the drying process (fig.4).

Fig.4 – Humidification to remove wrinkles and creases

Fig. 5 – Front, before treatment

Fig. 6 – Front, after wet cleaning and blocking

Adhesive Backing:

An adhesive backing of silk crepeline was prepared on silicone release Mylar™ by casting a 10% emulsion of a 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV adhesives to achieve the desired properties of not too tacky and not too hard. Once dried, the backing was laid out on a piece of blueboard covered in silicone release Mylar™ to prevent the backing from sticking unintentionally to any surface. The flag was placed on a Mylar™ strip while rolled up onto a tube and the realignment process began 8cm at a time.

The left side of the flag was lined up, on grain with the backing, with a T-square to ensure a straight edge. The flag was worked, bottom to top, in small sections, realigning the weave and matching any tears edge to edge. Once each area was realigned it was temporarily tacked in place with just the heat provided by a thumb. This allowed for a very weak bond which could also be lifted and shifted should the need arise. Once one section was done, the flag was lifted gently on its tube, the Mylar™ underneath repositioned 8cm further up the flag, and unrolled to expose the next section of flag. This process was repeated until the entire flag had been realigned.

An Optivisor was used to match distinct threads to ensure proper matching and alignment. As the silk has many irregularities in the size of the threads within the weave, it provided an inherent matching system which allowed for exact realignment. While the left edge was realigned straight, the distortion to the fabric was such that in some areas any attempt to realign the weave perfectly resulted in bubbles or stretching: the grain of the fabric had been permanently distorted. The decision was made to follow the new grain of the fabric, which resulted in all tears matching correctly and the flag lying flat on the backing without any bubbles or stress points. While the result is not a perfectly rectangular flag, it is properly aligned and hides the tears appropriately.

There are areas of loss; in these areas, what fabric remained was examined with use of the Optivisor and the remnant was placed in its proper position creating irregular areas of loss where some “islands” of fabric remain (fig.7).

Fig.7 – Detail of the areas of loss in the upper left corner

Once the flag was realigned, the adhesive was reactivated with acetone to create the strongest bond possible. The flag was still on the layers of blueboard and silicone release Mylar™, so the layering for the reactivation, bottom layer to top, was: silicone release Mylar™ -> adhesive backing -> textile -> Goretex™ -> blotter paper slightly dampened with acetone -> Mylar™ -> glass weight. Due to the previous dye bleed, precautions were taken to ensure only the required amount of acetone would come into contact with the flag. The acetone was applied to the blotter paper using a dropper to ensure it was not over saturated. During testing it was found that a 4-minute activation was optimal. This time was sufficient to totally reactivate the adhesive and created a good bond; no difference in bond was noticeable between 4-minute or longer activation times. The contact time was much longer during the poultice treatment with a stronger solvent. Still, this method was tested on the red sun before moving forward and found to not cause further dye bleed.

Small areas of 5x8cm (to match the size of the glass plates) were reactivated at a time. The next blotter was lined up with the first still in place, ensuring no areas were left without reactivation, nor were any areas reactivated twice. This process was repeated left to right, top to bottom, until the entire flag had been covered. Once this was completed, the flag was allowed to sit in the fume hood for 30 minutes to ensure any remaining solvent had evaporated.

Fig.8 – Front, after adhesive treatment

Fig.9 – Back, after adhesive treatment

Mounting:

A mount size of 38x30 inches was chosen to display the flag. The owner has chosen a stitch support without glazing at this point so he can have it professionally framed for display. This size will allow for an aesthetically pleasing border without any interference from the framer.

The strainer was assembled and adjusted with a T-square to ensure exact 90° angle. Once aligned, the corners were stapled on the back side to prevent further movement. Two fabrics, a cotton flannel and a mercerized cotton twill, were chosen as the support. Both fabrics were washed in a solution of 2% Orvus and deionized, pH balanced water and then rinsed 3 times with deionized, pH balanced water to remove whatever sizings or particulates might have been present. They were ironed to remove wrinkles, improve the hand, and perfect their appearance. A layer of white cotton flannel was stretched across the strainers on grain and tacked with thumbtacks on the rear side; this layer will provide enough padding and give to protect the textile when framed. The flannel was steamed and tightened to create a taut surface before being stapled into place on the back. Next, the same process was repeated with the show fabric, a blue-gray mercerized cotton twill, to create a visually pleasing appearance.

Once the mount was prepared the flag was aligned and centered to be visually pleasing. A layer of white 5mm silk habotai which closely matches the flag was cut with the use of a Mylar™ template and placed behind the flag to visually compensate for the areas of loss and to brighten the color. (An off white, to more closely match the flag, was suggested to the client but, after seeing both options, she indicated that she preferred the white.) The silk crepeline backing was too sheer to achieve this on its own and the blue-gray show fabric was visible through the areas of loss. The borders vary between 2-3 inches at different points to allow space for the hanging cords and the distorted shape. Once the flag was aligned, long and short stitches were used around the entire edge of the flag to distribute the weight and tension. The stitching was done through the silk habotai and mount layers to create even support and prevent further damage. It was stitched into place with hair silk thread and Number 15 beading needles which prevent it from being visually distracting on the front. The stitching was done through the interstices to prevent any further damage to the weave which had just been realigned and stabilized.

Assessment:

Overall, the treatment was successful and the goals have been met. The flag is much more stable than it was before the treatment. It was possible to mount it without causing damage to the fabric, which would have been nearly impossible before. It is also stable enough that it can be touched during the framing process without causing further loss. The tears are no longer overly visible, other than in the areas of loss. The dirt and stains have been drastically reduced and the appearance improved, allowing for display. It is mounted in a manner which improves both stability and appearance. Ideally, more of the dirt and staining would have been able to be removed, but this was impossible by any means which were safe or appropriate for the flag. Also, a backing of a color matching the flag would have been placed behind the areas of loss; however, when this was shown to the client she indicated that she preferred the white backing alone. The flag is in such fragile condition that this treatment should be considered permanent though it is technically reversible. Any further attempt to clean or re-back the flag will result in irreparable damage and loss to the fabric.

Recommendations:

The flag should be framed as soon as possible to prevent any further damage or buildup of dust or dirt on the surface. While any excess adhesive has been removed from the areas of loss, the adhesive will attract these particles through the loose weave of the fabric until they are prevented from doing so. In the meantime, the flag should be stored in an archival box to prevent possible damage or dirtying.

Fig.10 – Front, after treatment

Fig.11 – Back, after treatment

Fig.12 – Front, before treatment

Fig.13 – Back, after cleaning and adhesive treatment

APPENDIX 1: RESULTS OF TREATMENT TESTING

Solvent Testing Results:[1]

A piece of Silk Habotai which had been wiped around a mechanic’s shop to mimic the types of stains and particulates was tested with the following organic solvents. One drop of the solvent was placed onto one half of a cotton Q-Tip swab which had been removed from its applicator and placed directly on the fabric. The results were:

Ethanol: Direct application, 20 minutes, no movement
Methanol: Direct application, 20 minutes, no movement
Stoddard Solvent: Direct application, 20 minutes, no movement
Acetone: Direct application, 20 minutes, no movement
Ethyl Acetate: Direct application, 20 minutes, no movement
Napthol Spirits: Direct application, 20 minutes, no movement
m-Xylene: Direct application, 20 minutes, some movement

After the initial results, m-xylene was tested with the same method directly on the flag. It was found to move much of the soiling and staining and did not affect the dye.

M-xylene is an aromatic hydrocarbon with low polarity and the only solvent which moved any of the staining. As outlined in Appendix II of the treatment proposal, solvents from each of the main categories of solvents, and varying polarities, were tested. A wide approach was taken because the exact chemical nature of the soiling and staining was unknown; only that they were compounds which might be found in an auto shop. The ability of m-xylene to move the soiling and staining does confirm that they were greasy, oily deposits, but does not provide further insight into the exact compounds.

However, the fact that m-xylene was able to move the deposits while Stoddard Solvent, which also has low polarity but is the aliphatic hydrocarbon family, did not, does provide some insight. When both solvents were examined on the Teas Chart, it was found that the dipole-dipole, hydrogen, and Van der Waal’s values were actually quite similar. The values are:

m-xylene: N – 83, D – 5, W – 12

Stoddard Solvent: N – 90, D – 4, W – 6

The two solvents fall within neighboring quadrants. While this might be enough of a difference to mean that the m-xylene was more “like” the soiling and staining than the Stoddard Solvent, a more likely explanation comes in the form of molecular structure. Aliphatic hydrocarbons are aligned in a straight-chain where single, double and triple bonds are able to occur anywhere on the chain, resulting in localized electrons. Aromatic hydrocarbons contain a benzene ring and have alternating single and multiple bonds which allow the electrons to be delocalized. The benzene ring is a cyclic conjugated system which does not readily change configuration making it ideal for substitution reactions; in essence, it pushes other molecules out of the way rather than bonding to them. The straight-chain aliphatic hydrocarbons readily bond to many compounds. It is likely that the Stoddard Solvent was simply bonding with the particulates on the fabric rather than pushing the particulates off of the weave like aromatic hydrocarbons would tend to do.

[1] Due to availability restrictions testing with PERC, as outlined in the proposal, was impossible.

Poultice Testing Results:

During testing for lateral spread with agarose-water only mixtures it was found that 2% and 3% concentrations were too low to prevent significant lateral spread. The decision was made that 4% and 5% concentrations were most appropriate.

Fig.14 – Testing agarose concentrations for lateral spread

The poultice was then tested by soaking it in m-xylene overnight in a covered container in the fume hood to allow the m-xylene to diffuse into the gel replacing the water. This is an imperfect method still in development by Laura Mina and did not yield any results.

Next, a second method of brushing the m-xylene onto the bottom of the poultice directly before application to the textile was attempted and also did not yield any results.

After researching articles on poultices in both textile and paper conservation, the decision to try pure cotton pads in place of cotton linters was made, and was successful.[1]

It is likely that the concentration of solvent within the agarose poultice was not high enough to cause movement as a perfect recipe has yet to be developed. Normally, the additives are combined with the solution of agarose and water during the boiling process before it is set. Given the flammability of some solvents, including m-xylene, this method is inappropriate and dangerous. The cotton pads offered the direct contact with the textile, the capillary action to draw the soiling and staining out of the weave, and a medium to slow evaporation which would have been provided by agarose. However, using a dropper and cotton allowed precise knowledge of how much solvent was actually being applied to the textile, while the lab does not have the equipment to test the ratio of water to solvent within an agarose poultice.

Fig.15 – Testing with m-xylene and cotton, before

Fig.16 – Testing with m-xylene and cotton, during

Fig.17 – Testing with m-xylene and cotton, after

[1] See the bibliography on the treatment proposal for titles.

Adhesives Testing Results:

In order to discern the most appropriate backing fabric and adhesive, many options were prepared. All were tested on 5mm silk habotai which very closely resembles the weave and weight of the flag.

All adhesives are mixed in water and were applied in both the on-grain and bias directions to test for moiré patterning.

10% Jade 403N was cast on Stabiltex and reactivated with heat
20% Jade 403N was cast on Stabiltex and reactivated with heat
30% Jade 403N was cast on Stabiltex and reactivated with heat
40% Jade 403N was cast on Stabiltex and reactivated with heat
1.5% Klucel-G was cast on Stabiltex and reactivated with acetone
2.5% Klucel-G was cast on Stabiltex and reactivated with acetone

10% Jade 403N was cast on Silk Habotai and reactivated with heat
20% Jade 403N was cast on Silk Habotai and reactivated with heat
30% Jade 403N was cast on Silk Habotai and reactivated with heat
40% Jade 403N was cast on Silk Habotai and reactivated with heat
1.5% Klucel-G was cast on Silk Habotai and reactivated with acetone
2.5% Klucel-G was cast on Silk Habotai and reactivated with acetone

10% Jade 403N was cast on Silk Crepeline and reactivated with heat
20% Jade 403N was cast on Silk Crepeline and reactivated with heat
30% Jade 403N was cast on Silk Crepeline and reactivated with heat
40% Jade 403N was cast on Silk Crepeline and reactivated with heat
1.5% Klucel-G was cast on Silk Crepeline and reactivated with acetone
2.5% Klucel-G was cast on Silk Crepeline and reactivated with acetone

At this point it was decided that the silk crepeline in the on-grain direction had a very good look and hand and would be used. The silk habotai changed the color and the weave could be seen through the thin flag, which was visually distracting. The Stabiltex could also be seen through the flag and gave a very stiff hand. The silk crepeline blended in and was not visible through the flag; it also did not change the hand or create a ghost weave pattern. However, each of the adhesives had a different problem related to the desired look, feel, or bond. Further testing was done only on silk crepeline.

It was suggested by a fellow conservator that I should try a 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV. These adhesives are different molecular weights of thermoplastic acrylic resins. They would provide the temporary thermoplastic bond (using only the body heat of a finger) needed during realignment of the weave, but may also be reactivated by solvents rather than heat, which could be damaging to the flag. The adhesives are both archival and are shown to not yellow with age, giving them ideal properties for this project.

5% 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV was cast on Silk Crepeline and reactivated with heat
5% 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV was cast on Silk Crepeline and reactivated with acetone
10% 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV was cast on Silk Crepeline and reactivated with heat
10% 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV was cast on Silk Crepeline and reactivated with acetone

Silk crepeline with a 10% 1:1 mixture of Lascaux 360 HV and Lascaux 498 HV reactivated with acetone achieved the best result. This mixture of different weights of Lascaux provides the flexibility and elasticity of the 360 HV with the strength and ability to temporarily tack down an area with slight pressure of the 498 HV; the tackiness of the 498 HV is reduced to a good level by the addition of the 360 HV. The adhesive set without creating a stiff or rigid hand and without shine. The excess adhesive could be removed with acetone on a cotton swab from the backing which would show through the areas of loss.

Due to the ability for the solvent to penetrate into the fibers it creates the best bond and has, therefore, been chosen for this very fragile object.

Activation times ranging from 1 to 7 minutes were tested; at 4 minutes a very strong bond has been reached to the point that there are no noticeable differences between that and longer times.

Fig.18 – An example of adhesives testing

Fig.19 – Adhesives testing with Lascaux

A Stitch in Time

Textile Conservation