The Material Complexity of Three Seventeenth-Century Cabinets Exported from the Far East

ABSTRACT This paper focuses on three Far East cabinets that have ended up in Sweden for various reasons and on various routes. Some specific characteristics have been analysed using various scientific methods: sectional microscopy, wood taxonomy, X–ray fluorescence microscopy, pyrolysis-gas chromatography-mass spectrometry, strontium isotope ratio measurement, and radiocarbon dating. The results show that the cabinets contain wood from species endemic to Japan, namely asunaro (Thujopsis dolabrata) and hinoki (Chamaecyparis obtuse). They are coated with urushi sap derived from Toxicodendron vernicifluum trees, in two cases confirmed to be harvested in China, and one case probably blended with sap exuded from Toxicodendron succedanea harvested in North Vietnam. Their black appearance is due to a soot pigment typical of many products from the early Edo period or older. The data obtained with scientific methods, unavailable in the past, improve knowledge about these cabinets. Such new information should be made available to scholars and the general public.


Introduction
Historical trade with the Far East and the context of black urushi ware Before the sixteenth century, trade between Europe and East Asia relied on a land route, the so-called Silk Road, linking western China with the Mediterranean.After the discovery of the maritime route around Africa, the trading volume increased substantially.Early traders, i.e. the Portuguese, Dutch, and English, reached the coasts of Far Eastern countries and dominated the import of Indian, Chinese, and Japanese commodities.Tea, spices, porcelain ('china' in contemporary documents), and silk represented the most significant volumes of imported commodities.Researchers claim that the Portuguese were the main buyers in the early phase from the late sixteenth century, purchasing exotic goods with shapes familiar to their European customers (Okamoto 1972;Zetterholm 1985;Impey and Jörg 2004, 11-12;Nagashima 2008, 40).Until the expelling of Christians from Japan in 1613, European traders successfully made a profit (Kobayashi and Yoshida 2017, 17).During the next phase, in the 1630s, Japan was virtually closed to foreigners.Perhaps for that reason, and irrespective of their factual provenance, urushi ware from that period might be misinterpreted as Chinese.Some export urushi ware has been assigned a Chinese provenance, an opinion later questioned (Körber 2017, 35).For example, general cabinets with double-hinged doors frequently appear in export records from the mid-1630s (Nagashima 2008, 306).Although such a standard design does not alone point to a specific geographical origin, it may support the dating of the three cabinets.In the 1640s, only Dutch traders among Westerners were allowed into Japan and only into one southern port.It has been asserted that these Dutch merchants finally shifted actively from buying Japanese urushi ware to Chinese objects, such as large cabinets on stands, three or four decades later.However, at that time, the vogue for large cabinets had already declined and was out of fashion in Europe (Impey and Jörg 2004, 11-12).Narratives in historical texts tell us that these Dutch merchants regularly pressed for less-expensive wares, concerned about their profits once the imported goods were sold in European markets (Heginbotham and Schilling 2011, g).Documents from the Swedish East India companies occasionally also confirm that the prices of Japanese urushi ware were a concern for Swedish retailers in Canton (now Guangzhou).The Swedish traders wanted to profit from buying in Japan and selling at a higher price in China, so the margin for making a profit was at least partly dependent on the prices in Japan (SOIC Serie 3.4,Vol. 11).
Nevertheless, Japanese commodities are assumed to have been relatively scarce, due to ongoing trade restrictions, before the beginning of the Meiji restoration era in 1868.However, the style of urushi ware changed during the second phase (the 1630s-1640s) from the namban style with geometrical nacre inlay patterns to a more pictorial style, the so-called komo shikki, with makie in gold on a black coating.The combination of different decoration techniques; flat, raised, and burnished makie, or combinations of these, as well as the motives and styles of contoured cartouches and borders, has been used for dating of Japanese cabinets (Impey and Jörg 2004, 77, 86, 93-95;Nagashima 2008, 309;Otawa et al. 2016).

Description of the three cabinets
The three cabinets in this study are part of a survey of urushi ware extant in Sweden.This more extensive project, dealing mainly with Japanese urushi ware in public collections, has been conducted in collaboration between Meiji University in Tokyo and Uppsala University, Visby, over the last few years.The artefacts encompassed by the survey are mainly, but not only, categorised as namban or export ware (Brunskog and Miyakoshi 2020a, 2020b, 2022b, unpublished observations).Common denominators are these have been overlooked, out of context, or associated with unverified hypotheses.When the histories of museum objects are lost, it becomes a challenge to reconstruct them a posteriori.In the case of the three cabinets in focus, we decided to apply scientific analyses to learn more about their biographies.First, though, here is a brief summary of their current information.
Two cabinets belong to the collection of the Nationalmuseum, inventory number NMK 1/1921, henceforth referred to as 'NM1 ', and NMK 26/1965, henceforth referred to as 'NM26'.The third cabinet is in the collection at Skokloster Castle, inventory number 3088, henceforth referred to as 'Sko3088' (Figures 1-3).All three are of similar construction and design: a standard, squareshaped chest with double front doors enclosing many small drawers and compartments in five tiers.Their bodies are made of wood.They have an assumed Japanese provenance, as interpreted from entries in museum files and online acquisition catalogues.However, it is not always clear whether that provenance refers to the location of manufacture, the place of the export, an interpretation of style and artistic features, or some combination of these factors.Further, references to validated data are generally lacking.The same lack of underpinning data applies to the dates and materials of construction.
Cabinet NM1 is dated to the seventeenth century.Its exterior faces are burnished with raised decorations in gold, depicting children playing with birds in a rural setting (Figure 1).Its hardware, hinges, and corner fittings are of gilded and engraved metal.It includes a secondary four-legged stand painted black.It measures 75.5 × 87 × 51 cm (h × w × d).In 1921, Ester Lindahl bequeathed it to the Nationalmuseum to be displayed in her hometown of Nyköping (acquisition catalogue 1921).Before that date, its history is unknown.
The catalogue describes cabinet NM26 as made of Japanese black lacquer on a body of cedarwood (Figure 2).It is decorated with marquetry in nacre and lacquer paint in several hues: gold, cinnabar, green, light blue, and yellow-white, as well as with squarecut silver marquetry in dense patterns.The decor on the exterior depicts landscapes, trees, flowers, birds, male and female figures, a watercourse, a bridge, and a dragon.The decor inside the doors depicts flowers in a vase, one vase on each door.Plume-like pimpernels decorate the fronts of the drawers.The cabinet has fittings of engraved brass on all its corners, one large oval key escutcheon in halves on both doors, and three triangular fittings on the top.A separate, eighteenth-century leg stand was made by the cabinetmaker Georg Haupt.Cabinet NM26 measures (including the leg stand) 136.5 × 95.5 × 54 cm (h × w × d) (acquisition catalogue 1965).
Dating data for NM26, obtained from the museum, is highly inconsistent, comprising three different entries.In the acquisition catalogue, the entry date is 1600.On the other hand, in the online database, 1700-1740 is said to be the time of manufacture.The database states that the maker was an unknown Japanese craftsperson working in the seventeenth century.The Swedish princess Sophia Albertina (1753-1829) was one of the cabinet's previous owners.According to the wishes and the will of the bequeather, the cabinet is currently at the Foreign Ministry building, once the residence of the princess.In 1966, the cabinet underwent restoration in connection with its return, but to what extent is unknown to the authors.
According to the catalogue, cabinet Sko3088 is made of lacquered hardwood, possibly from a camphor tree (Cinnamomum camphora) (InsideWood), probably so named for its fragrance (Figure 3).Its exterior is a brown base colour with contoured or rectangular black areas.Rural landscapes are painted within frames, with birds and floral motives, a dragon, and a dog at the top.The decor is said to be partly done in ivory marquetry.The fronts of the drawers are done in black lacquer with floral motives, gold, and various dusky colours.The two doors have visible hinges, locks, and engraved fittings made of gilded brass.The feet are low contoured.Its measurements are 59 × 69 × 44 cm (h × w × d) (eMuseumPlus, n.d., online collection database), and dating data is inconsistent, with two different entries: the 1640s and the mid-seventeenth century.The latter timeframe has a broader scope of interpretation and possibly encompasses the former.This cabinet was passed on to the heir-of-law as a legacy.

Aims of the study
The aims of this study are twofold.The immediate one is to add to the information about these specific museum objects: three black cabinets showing similarities and differences.The rationale for investigating the three objects in the same project is to facilitate comparisons and to use the objects as references for each other, at the same time, to encompass variations if that is the case.Optical and chemical investigation methods can categorise and identify chemical components, describe coating systems, and underpin preliminary hypotheses on export urushi ware to enrich scholarship about the early export and import during the first globalisation era.In Sweden, few artefacts from that period have been closely investigated.Earlier scholarly research on multi-layered urushi ware aimed to answer questions about stratigraphy, and the chaîne opératoire for urushi production, presumed to show different trends in time and geography (Heginbotham and Schilling 2011;Ma et al. 2017, 123).The mapping of organic materials used in 'Asian lacquer formulations' is also a topic of the RAdICAL project (Getty Conservation Institute 2016) as in some other projects (Heginbotham et al. 2016;Schilling et al. 2016, S3).
Dating using scientific procedures can help date the cabinets in the study focus and others.Further, the results of the analyses might indirectly add to the knowledge of the deterioration of urushi coatings and, by extension, supply data for making decisions about preventive care or conservation (Reeves, Popelka-Filcoff, and Lenehan 2013).In addition, scientific analyses can complement and enhance culturehistorical aspects of urushi ware and its contexts (Archaeometria 2012).
Our research strategy has several goals: identifying the wood species used for the bodies, examining and characterising the coating system strata, determining the sap-producing wood species, and learning the geographic location of the trees exuding the sap(s) as well as the age of the coatings.Some other observations were outside the scope of this study: a comprehensive survey of the decoration techniques, weathering and signs of change, shell species, hardware, and secondary japanning.
The second, overall and enduring aim of this study is to promote the more-frequent use of archaeometry in the heritage field.In its broadest sense, archaeometry represents the interface between cultural history and the natural and physical sciences.Another aim is to exemplify the potential for archaeometry in a cross-disciplinary approach to collection management.Describing the biographies of objects and contextualising them are established ways for museums to curate their collections.However, most objects are anonymous or related with minimal context, perhaps, especially with urushi ware made for export.Hoping to gain more knowledge, experts from the West and East have joined to develop new information.
Early urushi export is often categorised as namban chikki, a convenient general term for works in heterogeneous but mostly over-decorated styles with a religious context.In Swedish museum collections, much urushi ware remains mis-or under-interpreted.As a result, objects can become neglected if people find them uninteresting.Objects with an 'appeal' are less at risk.Artefacts manufactured in Western styles, with locally obtainable materials and techniques, are easier to recognise for Westerners and may thus more easily evoke an appeal.Also, at the time of their acquisition, information about extraneous artefacts and their contexts might have been minimal, especially since many objects were inherited from generation to generation before they eventually became museum donations or purchases.During the last three decades, scholarly research on urushi has contributed new knowledge, and publications are accessible online, from which this study can also profit.

Sustainability as an important goal of collection management
Indirectly, the authors also wish to promote a widening of the concept of sustainable heritage as part of the reuse of objects and the development of collections management policies using multi-disciplinary teams.In our experience, cross-disciplinary studies are one, albeit not the only, way to produce competing histories (Nilsson and Garnert 2005).The sustainable management of collections should also mean the better use of already-existing resources, i.e. re-examining entire collections, specific categories of artefacts, and single objects, to gain more knowledge and profit more from their possessing.
Sustainability is often defined as meeting the needs of the present while at the same time not compromising the ability of future generations to meet their own needs.As a concept it is still open to interpretation, though, in a heritage context, 'sustainability' might be understood as the promotion of the repeated use of collections since these are the result of earlier and ongoing public investments, including acquisition, curation, safeguarding, and displaying.Improvements in energy efficiency in storage and exhibitions should always be strived for, of course, while the meaning of 'improve' has to be updated regularly with the most valuable guidelines.Using all other resources for maintenance, i.e. to increase the accessibility of collections and the information for which they are mediators, without putting them at risk, are other ways to practise sustainability.Further, raising the awareness of collections' multiple potentials, developing more comprehensible re-uses, and engaging a broader audience may increase the willingness of the public and stakeholders to allocate funding and pay taxes for their proper preservation by linking the social dimension of sustainability to accessibility.
Further, audiences may be involved as active spectators and developers of new meanings for museum objects and collections.For example, museums have invited audiences as partners by collecting memories and encouraging individuals to share their stories on topics like #metoo and #corona pandemic (nordiskamuseet.se,n.d.).Other museums provide a wide range of activities, from classes for primary-school children to residencies for professionals, that enable varied audiences to connect to and re-use collections (www.vam.ac.uk, n.d.).Of course, their intended reuses may be modified with such fragile antiquities and non-renewable museum objects.However, to remain meaningful, the meaning of objects has to be re-developed from time to time.Another strong argument for sustainable heritage conservation is that the fabric from which the built environment was  constructed should be preserved and kept usable as a waste-avoidance activity (Cassar 2009, 6).The same applies to artefacts, even if they represent only a fraction of the volume, and in addition, they might be made of material from extinct species.

Experimental
Our technical examinations followed an iterative approach, using standard scientific methods that required sampling for the characterisation of individual components in the bodies and coating layers of the objects.

Samples
Observation and macro photography were performed before the sampling.Samples were collected from undisturbed areas (i.e.without any visible restoration or other secondary material) and as discreetly as possible to honour ethical concernsthe latter favoured sampling from behind dismounted metal fittings and areas already displaying physical damage, i.e. cracking and loss of coating.The sampling process had to meet the need for many samples since some methods are destructive.Five samples were taken from NM1, five from NM26, and four from Sko3088.The specimens were distributed over most outside and inside surfaces except for exterior fronts and represent plain black urushi and areas with decoration.However, the full range of decoration techniques on each cabinet was too diverse to be covered (Figures 1-3).Similarly, raised makie (takamakie) were left uncovered since it was anticipated as an extra risky action, which might cause extensive material loss and unacceptably disturb the visual appearance.
The overall ambition was, for natural reasons, to answer the research questions: What components are there?How are they applied?When was it done?Sampling was done by covering areas with typical features of each specific cabinet, and in extension, displaying the general use of materials, primary techniques, and stratigraphy.Sampling for radiocarbon dating followed a slightly different ambition; in that case, plain urushi coating was suitable, obtained from the insides only, i.e. the backsides and insides of drawers.The larger samples were removed from the insides and backsides of the cabinets and contained only unadorned black coating layers.The samples are detailed in Table 1, showing the distribution of specimens for different types of analyses.Some samples included several fragments or were large enough to be subdivided during preparation in the laboratory.For pyrolysis gas-chromatography mass-spectroscopy (Py-GC-MS), 1 mg was required.Radiocarbon dating and strontium analysis required a minimum of 20-30 mg each, depending on the quality of the sample.Since the artefacts are of high culture-historical value, it is not easy to defend sampling of an amount of material that would be ideal for analysis.In practice, the analysis of the three cabinets had to comply with the obtained samples and not the other way around.
There was no separation of the 3-6 thin layers of 30-50 µm thickness each after sampling for Py-GC-MS and 14 C. Layer-by-layer sampling would probably have generated more detailed data (the Getty RAdICAL project; Heginbotham et al. 2016;Schilling et al. 2016;Tamburini et al. 2020).Regrettably, this was impossible to achieve in this study.However, the objective was to characterise the main constituents possibly present in the samples rather than to pinpoint each layer's exact composition.Probably, layer-specific data would not substantially influence the discussion on provenance and age but may add to material complexity.The intent was to remove only the original material, but secondary material was occasionally covered unintentionally.In such a case, the secondary material was separated before analysis as much as possible.
In most cases, though, observation of the samples with a digital microscope was the only measure before Py-GC-MS, and the samples were used without further pre-treatment.Similarly, the samples intended for 14 C analysis were used without pre-treatment to avoid the risk of contamination with modern material giving rise to a false radiocarbon age.The occurrence of incorrect age information was thus limited.

USB microscope photography
For close-up images of surfaces before sampling, a handheld Dino-Lite Digital Microscope Pro model in the AM/AD 411X series was used together with the DinoCapture 2.0 software supplied by the manufacturer.The microscope connects to a laptop computer with Microsoft XP software.

Stereoscopy photography
Before embedding, all samples were documented on both faces using a stereomicroscope Nikon SMZ1000 (Japan), Plan Apo 1x lens, and oculars C-W 10xA/22.A DinoEye eyepiece camera AM4023CT/R4 1.3 megapixel resolution captured the images.
Preparation of cross-and thin-sections and light microscopy (LM) Samples intended for epi-fluorescence observation were embedded in Epofix cold mounting epoxy resin, mixed with a hardener supplied by Struers.The epoxy cured overnight.Subsequent grinding and polishing were done on a manual Struers lab pol-5 using waterproof silicon carbide grinding papers with grits 220, 320, 500, 800, 1200, 2400, and 4000 stepwise with a few minutes on each grit.The Nikon Optiphot polarising microscope had an effective magnification of 50-400x.It was equipped with trinocular 10x eyepiece tubes, epi-bright/darkfield illuminator, halogen lamp, and plan achromat objectives 5x-40x.The microscope combines with reflected UV light for epi-fluorescence observation with a Nikon super highpressure mercury lamp model HB-10101AF.A Nikon DS-Fi2 camera and Nikon TV-lens C-0.6x captured the images, which subsequently were processed with Nikon Image Software Elements D version 4.20 for the best possible resolution, depth of focus, and undistorted images.
Samples intended for thin-section microscopy were embedded in epoxy resin, with any air bubbles removed with a vacuum and left to cure overnight.After subsequent grinding and polishing, the reverse side was glued to a microscope glass slide (Matsunami Glass Ind. Ltd.), using the same medium, and left to cure overnight.Thin-sections were cut to about 1-2 mm in thickness using a diamond saw (South Bay Technology, low-speed diamond wheel saw, model 650).Grinding was done stepwise automatically (Buehler AutoMet 250 EcoMet grinder polisher) for three minutes on each grid, using 400, 600, and 2400 mesh silicon carbide waterproof papers (for about one minute, the subsequent polishing was done with suspensions: Sankei aqra diamond suspension 3 microns, or Ultra-High-Purity Deagglom Alumina Suspen 0.05 µm).A final manual polishing step was performed to avoid excessive material loss and to reach a final thickness of around 10-15 µm.Observation of sections used an Eclipe LV100N POL microscope, Nikon Co. Ltd, equipped with a digital camera.Sections were observed in both reflected and transmitted light, in darkfield, and under crossed polarisers.

Wood taxonomy
From small wood splints in the samples, thin tissues were cut with a razor.All three principal directions of the wood were observed for tree species distinction with an Olympus Optical Microscope model BX51.Identification of wood species relies on their distinct anatomical features (Noshiro 2017, 25-26).The microfibril inclinations are the basis for forming the traits observed, but these are not used to identify tree species.The microfibril tilt angle is used as an index for evaluating materials.The hinoki cypress (Chamaecyparis obtusa) and asunaro (Thujopsis dolabrata) are distinguished by the form, size, and number of cross-field pits.The Chamaecyparis obtusa usually has two piceoid to cupressoid pits per cross-field (Noshiro, n.d.;2011, 129-131;2017, 25-26).Both hinoki and asunaro have exclusively uniseriate ray widths (IAWA Bulletin 1989, 219-332;InsideWood 2004onwards).Other scholars have used the microfibril angle to discriminate between these two species (Kita and Sugiyama 2021).

Pyrolysis-gas chromatograpy-mass spectrometry (Py-GC-MS)
Direct Py-GC-MS measurements were performed with a vertical microfurnace pyrolyser PY-2020iD (Frontier Lab, Japan), an HP 6890 gas chromatograph, and an HPG 5972A (Hewlett-Packard, Ltd.) mass spectrometer.A stainless steel capillary column (diameter 0.25 mm × 30 m) coated with 0.25 μm of Ultra Alloy PY-1 (100% methyl silicone) was used for the separation.A platinum cup with the sample (0.05 mg), was first kept on top of the pyrolyser at near ambient temperature; after that, the sample was introduced into the furnace at 500°C.The oven was programmed to provide a constant temperature increase of 12°C per min from 40°C to 320°C and held for 10 min at 320°C.The flow rate of the helium gas was 1 ml min-1.The injector had a split of 50:1.The MS ionisation energy was 70 eV (EI-mode).All pyrolysis products were identified by mass spectrometry at ionisation energy at 70 eV (EI-mode).Data were analysed with Agilent MSD Chemstation software (Idei et al. 2018, 1-5;Takahashi et al. 2018), and all pyrolysis products were identified from an interpretation of their mass spectra.The results are presented as extracted ion chromatograms (EIC).
X-ray fluorescence microscopy (XRF) XRF microscopy analysis was performed at normal atmospheric pressure with a current of 50 kV and 1.0 mA in the Horiba Scientific XGT-5200 analytical X-ray microscope with high spatial resolution, from 1.2 mm down to 10 µm.
Strontium isotope ratio measurement ( 87 Sr/ 86 Sr) A multiple collector-inductively coupled plasma-mass spectrometer (MC-ICP-MS), together with a Quadruple ICP-MS, was used to measure the ratio of strontium.Before measurement, the column was washed to eliminate residues.The sample was treated with 14 M nitric acid at 120°C for 11 h in a lidded Teflon beaker to dissolve the organic matter.Then, 5 ml of 14 M nitric acid, 0.5 ml perchloric acid, and 0.5 ml hydrogen peroxide were added to the covered beaker and continuously heated at 120°C for nine hours.After that, the lid was opened in dry heat to remove any organic matter.This operation was repeated until all organic matter was completely removed.Then, 1 ml of 7 M nitric acid was added to the sample and stirred to dissolve.From this solution, 1 ml was introduced into the column and eluted several times with nitric acid of different concentrations (Honda et al. 2015, 42).The application in this study used the lower ratio in Japanese soils, <0.710, compared with the ratio in Chinese soils with a threshold of >0.711 (Lu, Honda, and Miyakoshi 2012, 84-88;Yoshida, Sato, Nakai 2012, 13-27;Honda et al. 2015, 45;Yoshida 2017, 251-252).

Radiocarbon dating ( 14 C)
The Paleo Labo Co., Ltd, Gifu Prefecture, Japan, assisted with the radiocarbon dating.It was done by direct atom counting following the accelerator mass spectrometry (AMS) protocol (Tuniz et al. 1998).A Compact-AMS instrument, National Electrostatics Corp. (NEC), 1.5SDH, USA, was used.The machine was equipped with a dual ion source sequential injector, which allows measurements of low-activity carbon samples with masses 12, 13, and 14 for dating.The acceleration voltage was set at 500 kV.The precision is ±50 years, or, with careful tuning, even better.The curve calibration was according to the literature (Bronk Ramsay 2009;Reimer et al. 2013).All the dates are in conventional radiocarbon years before present (BP), with the year zero set to 1950 or the calculated common era (cal CE) unless otherwise stated.

Results and discussion
Table 2 summarises the most prominent results from observations and analyses in a condensed form.Some of the specific features are discussed in detail below.

Wood in bodies and drawers
The construction material is obvious, because wood is the first-choice body material for cabinets and many other sizeable constructions.The wood is partly recognisable because it has shrunk or distorted slightly from its plane, which shows through the coatings.The coating is also missing in some areas, leaving the wood bare.
The wood in both NM1 and NM26 has a typical conifer wood structure at a gross level (Figure 4 above and right).They display easily recognisable principal anatomical features, such as a more-or-less gradual change from earlywood to latewood within each annual growth ring (uneven grain), distinct growth ring boundaries, only one type of elongated fibre cells, and vessels are absent (IAWA 2007, 226).On a gross level, rays, pits, and potential resin canals are invisible.Erosion of the softer earlywood (inner zone of each annual growth) of NM1 is apparent, compared with the denser latewood (outer zone of each annual growth).However, the wood species remained unidentified from the macroscopic appearance.On ocular inspection, we only concluded the typical softwood traits of NM1 and NM26, together with the extensive abrasion from the drawers' repeated movement in NM1.The wood in the NM1 drawer is cut in the tangential and shows quite wide spacing of yearly growth.The same applies to NM26, where the wood is observable from inside the drawers' recesses, as shown in Figure 4 (right).The wood structure displays several macroscopically visible features, such as distinct growth ring boundaries and a gradual transition from earlywood to denser latewood.Knots are not present.The member in the right bottom tier recess is tangentially cut and has absorbed the blackening evenly, irrespective of late-and earlywood.(The area closest to the opening looks darker due to an unidentified film-forming agent, probably an urushi application.)These observations support the interpretation of a conifer species (InsideWood 2004 onwards).The same macroscopic features likely also apply to Sko3088, although the bare wood area was much smaller and less easily observed compared with NM1 and NM26.Still, long fibrils and a typical shift in hue between early-and latewood were observable.
Although growth-ring boundaries seemed wavy on a macroscopic level, this was not confirmed by the samples prepared as sections where they appear straight (Figure 4).Undoubtedly, skilled joiners of yesteryear understood the best practice of cutting wood radially into dimensionally stable boards that would not warp, cup, or twist.In contemporary Europe, most lumber would not be sawn lengthwise but split with a wedge, thereby attaining a growth ring pattern orthogonal to the board faces, i.e. vertical grain.The same quality concern applies to China and Japan (Heckmann 2002, 79).However, wedges of wood were used far less in traditional Japanese carpentry than would be the case for equivalent tasks in  the traditional crafts in the West.The Japanese Edo-era ripsaw (tatebiki-noko) was shaped to cut along the wood grain (Takenaka Carpentry Tools Museum 2010).Usually, wood was and still is, seasoned for several years.The grain structure, wood orientation, and joints are carefully thought out and meticulously executed.Later, when sawmills became common, the same was attained by perfectly quartersawn grain, the most prized grain pattern (Diderot andd'Alembert 1751-1765;Roubo 1772).Species-level identification is indispensable for clarifying the provenance and historical use of timber.Since there are several anatomically similar wood species, microscopic analysis was necessary and available.Cypressaceae trees were evaluated for their straight grains and durability and, therefore, frequently used throughout Japan in historic times.As stated by Noshiro (2011, 125), species-level identification of species of Cupressaceae can be based on quantitative analysis of variations in cross-field pitting and rays.The size, type, and frequency of cross-field pitting allowed for identification in this study.
Microscopic data confirms that the body in NM1 is asunaro (Thujopsis dolabrata BUN 1249) (Figure 4 above).The form and frequency of cross-field pits distinguish asunaro from hinoki (Chamaecyparis obtuse [Siebold & Zucc] Endl.) (Noshiro, n.d.;2011, 126-129).Asunaro and hinoki share some cross-field characteristics: pits show circular borders, aperture openings tilt, and the number of pits per cross-field.However, they appear different in other respects: asunaro has cupressoid to taxodioid pits, with a pit diameter of less than 5.5 µm that occupy various places of the cross-fields.In comparison, hinoki has piceoid to cupressoid pits, with a pit diameter ranging from 5.5 to 6.5 µm placed in the vertical centre (Noshiro 2011, 128, 129, Fig. 1 a-d, and Fig. 2 a-d).
Asunaro is endemic to Japan.Traditionally it was valued for its scent and durable wood and was planted around temples and in gardens.As it happened, Carl Peter Thunberg, a Swedish physician, botanist, and zoologist, travelling Japan in 1775-1776, was the first to describe this wood species characteristic.At the gross level, the wood in NM26 (the sample photographed before embedding) appeared so similar to the wood in NM1 that it was decided that preparing a thinsection dedicated to wood taxonomy was unnecessary (Figure 4), also considering the limited access to resampling.In an ideal situation, removing wood tissue from NM26 would have made a more trustworthy identification possible, but the restrictions during sampling on-site made this unachievable.The conclusion is that the bodies in NM1 and NM26 might be of the same wood species.At least, the claim on cedarwood in NM26 is convincingly refuted.Plants possess terpene compounds, including mono-and sesquiterpenes.Since these have a pleasant scent, they often serve as raw materials for making cosmetics, perfumes, and soaps.So, even if there is no cedarwood in NM26, aromatic vapours could have been emitted from the carcase of the cabinet, partly explaining why the catalogue text mentions a fragrance.Non-anatomical features such as odour may have volatilised from the surface of old wood.Odour is also quite variable, and individual perception often differs, so any indication should be interpreted cautiously.
The wood from one of the recess dividers in cabinet Sko3088 was identified as Cupressaceae Chamaecyparis obtusa (Siebold & Zucc.)Endl., endemic to Japan and commonly called hinoki in Japanese (Figure 4 below) (Noshiro, n.d.).The cypress native to Taiwan is a variety of Japanese cypress.Cell lumina visible in one of the thin-sections have a rectangular shape which might reflect that the cells are from the narrow latewood zone only (Figure 11 below).Usually, this dense zone is narrow compared with the earlywood zone.Rays are typically a single cell-wide (InsideWood 2004onwards; Farjon 2015; the Wonders of Urushi 2017, 22), and the appearance and frequency of pits in cross-fields are used to differentiate hinoki from cypress.The cross-field pitting is seen on the tracheid radial walls (Figures 4, R-cut) (Noshiro 2011).Hinoki has been valued as a construction material since ancient times and is associated with its fine grain, strength (and resilience), and workability.The wood cannot explain the fragrance of camphor in Sko3088; nevertheless, hinoki is appreciated for its long-lasting and lemon-like scent.Fresh wood displays a pinkish heartwood that is distinct from sapwood.With ancient samples, wood colour cannot be used since the wood is altered by age, use, and deterioration.The International Union for Conservation of Nature (IUCN) categorises hinoki as globally 'near threatened', corresponding to four on a nine-grade scale, as assessed in 2011 (Farjon 2015).
Both kinds of wood confirmed or assumed in the three cabinets are traditionally associated with making urushi ware in Japan since the Heian period.Hinoki was used in the still oldest extant wooden building in the world, the five-storied padoga in Horyji temple, Nara Prefecture, designated as a UNESCO World Heritage site.Dendrochronological dating indicate that the pagoda is contemporary with the Yayoi and Asuka periods (BCE 241-CE 591) (Index Xylariorum 4.1).The use of locally obtained wood is assumed to be the general rule, with few exceptions.The same rule is also reflected in the collection of John James Quin, Kew Gardens, London (Heckmann 2002, 81, 88).Previous studies on export urushi ware in Portugal have shown that hinoki was used.Thus, the artefacts were interpreted as Japanese (Oka 2017).Heginbotham et al. (2016, S3-73) has confirmed hardwood in Chinese objects.Other possible species include Japanese Thuja (Thuja standishii), Powlonia tomentosa C.
Koch, Chinese juniper (Juniperis chinensis), and Japanese cedar (Cryptomeria japonica) as well as Cupressus, Juniperus, and China fir that grows in China (Project 2001, 329;Noshiro 2017).However, none of these was confirmed in the examined cabinets.
The scope of this study is limited.Thus, it does not explain why some bare wood in NM1 and NM26 appears tangentially cut.If future research includes a CT scanner large enough to scan cabinets, it may shed more light on the issue.Export lacquer cabinets in the collection of the Staatlichen Münzsammlung, Munich, have shown that care and attention to detail did not always apply, with boards cut across the grain (Heckmann 2002, 79; www.museum.com/jb/showdia?id=4624).

Modus operandi of the coating process
In general, the urushi-making traditions have been almost the same regardless of temporal and spatial differences, varied workshop routines, and trends in styles when considering primary structure and sequence of layers, with their different functions, from bottom to top, i.e. what type of layer followed after another.Nevertheless, at the same time, there is an almost infinite variety of techniques and materials (Heckmann 2002, 124).The edge of a split in the coating on the left-hand door's exterior face on NM26 displays a typical coating structure (Figure 5).The cleavage is perpendicular to the surface, and the cupping is parallel to the wood.Therefore, the edge of the detached coating displays the various upper layers, including the decoration and transparent urushi layers.Beneath the disjoined layers, some fibres are visible on a beige or beige-reddish urushi ground (Körber 2017, 35).
At the same time, and, in general, some details in stratigraphy and composition may be highly complex (Nagashima 2008, 68;Heginbotham and Schilling 2011;Heginbotham et al. 2016, S3-36;Hao et al. 2017;Sato 2017, 85;Matsuda 2019).The number of applications of each layer, the main components of fillers, film-forming agents, and different additives may vary widely, as well as the total number of layers.Sometimes a textile cloth layer was and is still used at the interface between the wood substrate and actual ground (International Course 2001, 36, 76;Heckmann 2002, 218-219, 227;Lu and Miyakoshi 2015, 268;Honda 2017, 67; the Wonders of Urushi 2017, 86).Such a base for the actual urushi ground layers is known as nunokise in Japanese (Nakasato 1993, 70;Heckmann 2002, 84-85).In other practices, a paper substitutes for the fabric, occasionally also between ground layers (Brunskog 2003, vol. II GSM 20.788;Nagashima 2014, 20;Hao et al. 2017, 4;Brunskog and Miyakoshi 2020b, unpublished observations).The intention was probably the same, to mitigate stress, reinforce vulnerable points, bridge gaps, and provide a flat and homogenous surface.The use of textile is assumed to be more sophisticated than paper (Kawada 1993, 29;Nakasato 1993, 70;Heckmann 2002, 68).This view on the issue of sophistication is contradicted by findings on a Chinese imperial coffin (Hao et al. 2017, 4), where the coating structure has 10 layers of paper and stucco in alternating order.
The section NM1-2 sample (Figure 6), captured under moderate magnification, displays a four-layer structure.A paper may exist between applications in the lower parts of the ground (d).The same figure shows the areas within the white rectangles enlarged.The shapes resemble flat fibres with narrow lumina similar to those observed in other cross-sections (unpublished observations) and are interpreted as ramie.In this case, the fibre species was impossible to determine.Py-GC-MS analysis of NM1 confirmed a polycarbohydrate compound (Figure 16).Perhaps further research will help reveal whether the detected carbonates are pyrolysis products of a paper, textile fibre, or some other substance in the same material class.Under moderate magnification, fibres were not observed in the thin-sections of NM1-3 B, C, and D (Figure 7).However, the grounds are responsible for more than half the thickness compared with the total coating thickness.Since the sections in Figure 7 display no wood, it is uncertain whether all ground layers were captured in the samples.Under higher magnification, fibres were observed in NM1 fragment C (Figure 8).The fibres appeared soaked with urushi sap before the subsequent urushi-soot mixture was applied.The fibre diameter is medium, with an average width of c.50 µm, flat, and with a wide lumen (Bergfjord and Holst 2010, 1192-1197;Angelini and Tavarini 2013, 138;Bunsell 2018, 62, 312, 317).The NM1-2 fragment, viewed under higher magnification, also shows a more elaborate coating, displaying a seven-layer structure (Figure 9).This thin-section shows a black layer corresponding to layer b in the other two fragments, B and D, thicker and of a different structure, not observed in the other two thin-sections.Moreover, several diverse applications were observed on top of the translucent urushi application: an urushi-soot mixture and applications containing cinnabar (maybe silver), gold, and transparent urushi, respectively.The gold was both granular and in flakes.XRF confirmed the metals Au (gold) in high concentration and Cu (copper) in lower concentration, indicating gold of a lower carat.Other detected elements included Fe (iron), Ca (calcium), Si (silica), and K (potassium), interpreted as minerals in the ground (Figure 12).Cabinet NM1 has many types of sprinkled, inlaid (kirigane), and makie decorations (Figure 1), including flat relief (hiramakie, hira = flat) and raised relief (takamakie, taka = high), of which the latter explains the many layers applied after the clear urushi, captured in this section.
In the unprepared NM26 sample No. 4, there seems to be a separation zone between the ground layer and  the transparent coating (Figure 10 above).The thinsection NM26 (Figure 10 below) displays fibres below the ground layer e and cell cavities visible in both reflected and transmitted light.The fibres do not look arranged as they would be in a piece of fabric but are more randomly distributed, as in a paper.They might be ramie, as interpreted from their flat appearance, but hemp cannot be ruled out without further analysis.
There is also an indication that paper was used to make Sko3088 (Figure 11 above), but it was not apparent from another section sampled from the same cabinet (Figure 11 below).The cross-sections of the fibres are not as flat as those observed in Figure 8, leading to the assumption that they are of a different species.The exact positions were perhaps not ideal for capturing fabric or paper in all samples.A sample covering mainly restoration material also contained a paper-like material.The original ground layers likely remain under these retouched areas, including a potential paper covering the entire surface or specific vulnerable areas only: corners, rims, and edges.
The ground layers on NM1 and NM26 are mixtures of fillers and urushi, appearing opaque, homogenous, and relatively thick compared to the total coating thickness.The fillers are coarser and finer particles of different colours and irregular shapes but mostly greyish-beige-reddish, as expected from a natural compound such as clay.Since the ground layers look similar from bottom to top, it is impossible to tell whether they result from a sequence of applications, as expected, or just applied once on NM1 and NM26.In other studies, urushi grounds appear in many sequences, as mentioned above (Brunskog and Miyakoshi 2020b, 9-10).Generally, a layered structure can be seen more clearly if there is a shift in colour or structure.The variations can be explained in many ways: e.g.differences in composition, insufficient adhesion, a long enough time between applications, or a surface once exposed to dirt.The cross-sections, shown in both visible and ultraviolet light and under crossed polarisers, confirm a mineral content in the grounds with irregular, whitish, yellowish particles.XRF of NM1 and NM26 confirmed the presence of iron (Fe) in a high concentration in NM26 in combination with some silica (Si) (Figure 12).
Traditionally, many different types of clay have been used to provide a buffering zone between the substrate and the lacquer layers.The filler is either kilned jinoko or washed tonoko, as the main ground layer constituent.Different recipes prescribed mixing the clay with either urushi, paste, glue, or a combination of those.However, other filler materials are not expected (Heckmann 2002, 83).Thus, iron or iron ions in urushi coatings may usually be present as minerals in the ground.Iron compounds can also be detected and explained as a colourant of a clear coating to render a deep black colour or a red undercoating for makie powder or metal leaf.The fluorescent X-ray data of NM1 and NM26 are interpreted as clay fillers such as jinoko or tonoko, also based on the observation of transparent urushi layers.The yellow-reddish hue of the ground layers observed in the cross-sections indicates a jinoko rather than tonoko, but both types are traditional materials in urushi grounds (Egashira and Ichikawa 1999).A darker-coloured ground may indicate a higher proportion of urushi in contrast with a lighter-coloured one, which may be due to a higher  proportion of other binders or a combination of urushi and other binders (Heckmann 2002, 82).
On both NM1 and NM26, there is a thin blackcoloured layer on top of the ground observed in the thin-sections.On Sko3088 sample No. 1, this black layer seems to be applied directly onto the wood substrate (Figure 12).It is also much thicker than on both NM1 and NM26 (Figures 6-7, 9-10).Thus, the coating structure observed on Sko3088 sample No. 1 appears different and less elaborated than NM1 and NM26 since a thick ground layer, and fibre reinforcement are sometimes missing.The thick black layer directly on the wood substrate contains single, larger translucent particles interpreted as silica and many minute, rounded, black particles or agglomeratesthe pigment.It has penetrated the outermost wood cell layers, considerably improving the physical adhesion by interlocking.The black layer is thicker (estimated 50-60 µm) than the transparent urushi (estimated to be 20 µm) applications.The effect of this structure is assumed to render the cabinet a deep black colour due to the penetration, absorption, and reflection of light.Section Sko3088-2 indicates that a thin ground layer is present in some areas, and single fibre fragments were observed (Figure 11 above).
XRF detected gold (Au) in a sample from Sko3088, which may explain the brown colour noted in the museum file.Light reflected from the surfaces of gold flakes passing through a transparent urushi coating can create such a colour.
Py-GC-MS EIC at m/z 202 data from all three cabinets confirm that retene and pyrene were detected.In the EIC from NM1 and NM26, the retene peak was of lower relative intensity than pyrene, and vice versa, the EIC of Sko3088 (Figure 13).These types of pyrolysis products confirm carbon black (soot).Charcoal (wood ash) is one of the oldest pigments used in pre-historic times.Soot is a known method of making black urushi, traditional before the mid-eighteenth century in China and Japan.

The botanical and geographical origins of the saps
The natural polymers, tapped from three tree species in the Anacardiaceae family, have been used in East-Asian coating technology for millennia.The polymers are exuded from either the Toxicodendron vernicifluum (urushiol) that grows widely in Japan, China, and Korea; the Toxicodendron succedanea (laccol) that grow mainly in Vietnam and Taiwan; or the Gluta usitata (thitsiol) that grows mainly in Thailand, Lao, Cambodia, and Myanmar (Burma).However, Taiwan is a relatively new region for the plantation of T. succedanea which started around 1925.In 1986, a typhoon caused damage whereafter the sap harvest ceased, suggesting that coatings older than that period were made from Vietnamese yields (Nagase and Miyakoshi 1998, 31-36;personal communication Hsu, n.d.).(The Yu-Fu Hsu family used to plant lacquer trees in Nantou County, Taiwan, but now runs a lacquer museum.)Different species have different film-forming components, and each yield is affected by local and temporal growth conditions and tapping season and technique for example (Lu and Miyakoshi 2015, 2-9).Depending on how these are used (individually or in blends), the cured coating-film composition becomes different (Anzai et al. 2014, 139-140).Ageing during exposure to air and light and heating during Py-GC-MS analyses also affect and decompose the chemical structure (Kamiya et al. 2006(Kamiya et al. , 1311;;McSharry et al. 2007, 34-36;Lu and Miyakoshi 2015, 220).Therefore, the thermally degraded fragments of such coatings detected in Py-GC-MS are never the same as the components in the raw sap, but these are inferred from the vapours, the pyrolysates since they produce characteristic pyrolysis products (Ma, Lu, and Miyakoshi 2014, 137;Lu and Miyakoshi 2015, 223).The main sap components take on a wide range of bonding structures; thermal degradation makes some of these components revert to the original and not others.When a coating film is thus pyrolysed into fragments of various shapes, chemical rules dictate that several kinds of structures tend to emerge the most.When a sap-based coating is analysed, a mass fragment m/z 108 from alkyl phenol and a mass fragment m/z 123 from alkyl catechol appear as characteristic fragment ions (Lu and Miyakoshi 2015, 224).In the case of Toxicodendron vernicifluum, the mainly detected components are 3-heptylphenol and 3-pentadecylcatechol.By contrast, the most distinct components of Toxicodendron succedanea are 3-nonylphenol and 3-heptadecylphenol (Honda 2017, 59).Therefore, it becomes possible to detect and differentiate sap-based coatings by analysis (Honda and Miyakoshi 2012, 232;Lu, Honda, and Miyakoshi 2012, Tab. 3-1).Once the main components are identified, the tree species and indigenous region can be deduced (Ma, Lu, and Miyakoshi 2014, 141).
The raw sap used for the three cabinets' fabrication was exuded from the same tree species, the lacquer tree Toxicodendron vernicifluum.In the Py-GC-MS EIC at m/z 108, the peaks C7 and C15 indicate that the three separate coatings are chemically similar (Figure 14 above).Urushiol is a series of catechols, of which the most abundant pyrolysate has seven carbon atoms (3-heptylphenol) and shows the highest EIC relative intensity.The member with the maximum chain length has fifteen carbon atoms in the side chain (3-pentadecylphenol).The result holds for NM1 sample collected from the interior and coating samples from NM26 and Sko3088.
However, unexpectedly, NM1 sample collected from the exterior has a different composition of saps indicated in the EIC (Figure 14 below).The peaks C9 and C17 are markers for compounds associated with pyrolysis products from laccol, namely 3-nonylphenol and 3-heptadecylphenol.
Such sap (laccol) was exuded from Toxicodendron succedanea, distributed in North Vietnam (Lu and Miyakoshi 2015, 231-233, 237;Honda 2017, 59-61).It would seem logical to use less-expensive raw materials on the inside where the aesthetics and finish might be less critical if prices need to be kept low, as discussed above (Heginbotham and Schilling 2011, g).Other reasons to mix saps of different botanical and geographical origins might be to modify the drying properties, flexibility, and gloss.Laccol saps dry more slowly and becomes lighter in colour after curing than urushiol (Honda et al. 2008, 75).The peaks' relative intensities in the EIC indicate that the laccol was the primary ingredient, i.e. the laccol was modified with urushiol instead of vice versa.We hypothesised that the artisans had the experience, supply, and skills to choose the appropriate material for different applications and use them correctly.Therefore, we assume that a lighter-coloured urushi sap would better enhance the lustre and colour of the gold powders and flakes, which are densely distributed on  the cabinet's exterior.A similar blend of saps containing indigo was confirmed on a desk with Japanese urushi panels (Brunskog and Miyakoshi 2020b).The final look of the NM1 cabinet was probably the primary concern for the artisans.Besides, if the time for curing were an issue it would be best practice to hasten the process with an addition of, e.g.urushiol or other alternative substances (Matsuda 2019, 85-86).Deduced from analysis findings of excavated urushi ware in Kyoto, it is reasonable to conclude that mixtures of saps were also used on other artefacts, aimed at the domestic market, during the Momoyama era (Honda et al. 2010;Kitano et al. 2014).Objects from this period include a wide variety of commodities for the domestic market; utensils for personal use and costume accessories, utensils for the tea ceremony, as well as different kinds of containers, cups, bowls, jars, and pots for daily use in households, but also artefact aimed for export (Honda et al. 2010).
Furthermore, about the excavated artefacts and documentary evidence, Kitano concludes that the imported sap traded at Hirado Dutch trading posts was more expensive than domestic sap and used during the 1630-1640s perhaps for relatively highquality products (Kitano et al. 2014, Table 1, 74, 77).On the other hand, the price of urushi seems always to have been high compared with other commodities (Thunberg 1791, 159-160;Nishikawa 1993, 23;Heckmann 2002, 14, 82).The effort to save money also seems to have been a factor, as suggested by archival sources from mid-fifteenth century Kyoto and 'how people of those days racked their brain to reduce costs' (Kaji 2018).Hitherto, only a few examples of coatings based on blends of different anacard saps have been reported.Single items belonging to the Urasoe Art Museum, Naha, Okinawa, are made from urushiol and laccol (Honda et al. 2015, 41-45;Miyakoshi, Miyazato, and Honda 2021).The same applies to Japanese panels attached to a Rococo writing table in the Royal Swedish household (Brunskog and Miyakoshi 2020b) and five artefacts reported by Kitano et al. (2014), comparable with the cabinet NM1.Findings of thitsiol in admixture with urushiol have been detected on four eighteenth-century pieces of furniture with Japanese lacquer panels in the Getty Museum's collection (Heginbotham and Schilling 2011, f).Except for the latter, these limited examples have pigment colours (green, blue, gold) that probably look more vivid when mixed with a less stained sap.More research is needed to clarify the exact circumstances surrounding when and why sap mixtures were preferred if they were feasible.Still, it is outside the scope of this study.Such a study would require a statistically sufficient number of artefacts with known context, characteristics, and age that are feasible to sample.Suffice it to say: that the issues of pricing and artistic routine are multifaceted.
Strontium isotope analysis data indicates that the saps used on NM1 and NM26 were harvested in continental Asia (Figure 15).Combined with the saps' main constituents, this points to origins in mainland China and North Vietnam.The strontium ratio for NM1 and NM26 is 0.7235 and 0.7134, respectively, above the threshold value.The threshold between mainland China and other continental regions, and Japan is 0.711, marked with a grey zone in Figure 15.The sample volume from Sko3088 was insufficient to allow strontium isotope analysis, making it impossible to either confirm or reject hypotheses on the provenance of the urushi sap based on it.
Previous research has confirmed that the trade of urushi sap was extensive between countries and regions in the Far East and is well-known from historical documents (Kitano et al. 2008(Kitano et al. , 2009(Kitano et al. , 2014;;Kobayashi and Yoshida 2017).Moreover, containers with raw urushi, excavated at sites in Kyoto, have been analysed, and compounds detected that are markers for saps from Myanmar (Burma), Cambodia, Thailand, and China.The results testify that all three types of urushi sap were used in Kyoto during the Momoyama period .The coastal cities of Nagasaki and Hirado in Japan are believed to have been the trading ports for saps from other regions in Southeast Asia.The increase in imports to Japan can partly be explained by the rapidly growing export of urushi ware and the simultaneous construction of many temples and shrines throughout the Japanese empire.It is assumed that imported saps were used mainly for goods for the West.The demand for imports increased to the extent that domestically harvested supplies were insufficient.The workability and final properties were also modified by blending saps of different origins.Mixtures of saps from different tree species grown in different areas are ways to modify viscosity, drying speed, gloss, and other final physical properties.(Heckmann 2002, 16-18;Honda et al. 2008).Thai sap requires a higher temperature (≥30°C ) and humidity (≥80% RH) to set compared with Chinese and Japanese saps (20-25°C, 70-75% RH).The latter would shrink and develop wrinkles if exposed to such an environment.Astonishingly, the artists could master the process without the help of thermometers and hygrometers.Py-GC-MS and FTIR data suggest that the raw saps in containers excavated in Kyoto were used individually but also in different blends.(Kitano et al. 2008(Kitano et al. , 2009(Kitano et al. , 2014;;Honda et al. 2010).The four types of composition detected are sap from Gluta usitata (thitsiol) singly and mixed with either Toxicodendron succedaneum (laccol) or Toxicodendron vernicifluum (urushiol) as well as Toxicodendron vernicifluum (urushiol) singly (Kitano et al. 2009).These mixtures correspond to the findings referred to above.

Additives and other components in or on the coating layers
Other compounds detected in Py-GC-MS include cedrol, dammarane, alpha-cedrol, alpha-cedrene, and 3,4-secodammarine in samples from NM1, and alphacedrol and alpha-cedrene in samples from Sko3088.These compounds are volatile or semi-volatile, might cause a pleasant scent and thus may partly explain the noted fragrance, referred to above, believed to indicate cedarwood in NM1 and camphorwood in Sko3088.Camphor derived from the camphor tree (Cinnamomum camphora L.) is a waxy solid or oil used as a diluent or plasticiser.Therefore, a potential additive cannot be rejected; however, its presence in urushi coatings has been confirmed in only a few instances (Heginbotham et al. 2016, S3-5) and not at all in this examination.Py-GC-MS data are not layerspecific, so it is impossible to know from where the detected compounds are emitted.
EIC data at m/z 60 from all three cabinets confirm the pyrolysates palmitic acid (C16) and stearic acid (C18), and the peaks C6-C9 mark pyrolysis products for other related saturated fatty acids.Figure 16 shows the data from NM1.The same intensities also apply to samples from NM26 and Sko3088.The P/S ratio is relatively higher in all cases, but the difference between the P and S is less in the sample from Sko3088.These pyrolysates can stem from several organic substances, such as fat, wax, and oil, and suggest the presence of lipid material (Mills and White 2011, Chap. 3).The same result has been obtained in other analyses of Japanese urushiware of similar age (Brunskog and Miyakoshi 2020b, 15;2022a, Fig. 15;2022b, 7).
Anacard saps contain small amounts of polysaccharides, glycoproteins, and laccase enzymes in phenolic lipids such as urushiol, laccol, or thitsiol as the main component.However, neither palmitic, stearic acids nor their esters are native components of saps nor any other fatty acids (Honda et al. 2016;Sung et al. 2016;Hao et al. 2017, 6 and Figure 8a;Idei et al. 2018).
In contrast to unsaturated fatty acids, saturated fatty acids (such as palmitic and stearic acids) are chemically stable and not involved in the sap's hardening (Wang et al. 2015).Thus, they are considered to remain in the coating film without being oxidised.It is believed that they are thermally decomposed by Py-GC-MS analysis and hence possible to detect (Honda 2017, 60).
Several oils contain palmitic and stearic acids (Mills and White 2011, 36;Wang et al. 2015).Therefore, these alone are not conclusive evidence for the presence of a specific oil.The detection of azelaic acid and other dicarboxylic acids would provide such evidence.However, with what is empirically known about the Chinese, Vietnamese, and Japanese urushi craft traditions, adding oil was indeed a conventional procedure.Again, it is essential to note that due to the petite size of the analysed samples with multiple lacquer layers, when present, they were not isolated.They were analysed as such; therefore, the results cannot indicate in which layer a specific lipid component is contained.This study's results are inconclusive and in the absence of irrefutable datawe suggest further analysis before any conclusions are drawn.
In the same EIC, the peak AGP confirms the decomposition products of starchy material (Figure 16).The peak indirectly indicates, for example, plant fibres or flour of unspecific origin.The sections clearly show that fibres were present in many samples.Starchy materials have been mixed with sap, e.g.kiritsukekozane (urushi and flour paste), mugi urushi (ki urushi and wheat flour), and nori urushi (ki urushi and rice paste adhesive) (Project 2001, 222-223;Glossary of Urushi Terms 2005, 9, 12, 14).Traditionally, wheat or rice paste was used to increase the stickiness of an urushi film and extend it, thus reducing the consumption of sap (Heginbotham and Schilling 2011, Table 1; Encyclopedia of Japan 2012, 324).Recipes for grounds occasionally also prescribe the urushi to be extended with rice paste as a binder for the filler, as mentioned above.This study's observation of starchy materials is perhaps consistent with assumptions about export ware as generally lower quality than artefacts aimed for the domestic market.Alternatively, at the least, it supports the hypothesis that saps were expensive and required cost-reducing measures whenever possible.It seems logical that this depended on whether the sap was harvested locally or overseas.The result cannot differentiate in which layer(s) the polycarbonate is present; at the same time, it confirms that fibres make up at least part of the carbonates.In other studies, the foundation, lower, and upper finish layers have been shown to contain starch (Heginbotham and Schilling 2011, Table 1).However, in this study, the circumstances did not allow for the separation of layers during sampling, a prerequisite for layer-by-layer analysis, as reported by the Getty RAdICAL project.
Compounds with amine groups (Figure 17) were detected in samples from NM26 and Sko3088, respectively.Amines indicate a proteinaceous material, for instance, animal glue, traditionally used as a binder for metal foil in decorations or as additives in lacquer film (Hao et al. 2017, 6).Based on the analysis, we assume that animal glue has adhered to the metal decoration.

Dating of the coatings
A 70-80% probability in radiocarbon dating is considered a strong indication, while 95% is even firmer.The error range is narrower or broader depending on the standard deviation (σ).The more limited suggested periods of manufacture of these cabinets have the boundaries set to a 68.2% confidence range, or one standard deviation (1σ).The less limited suggested periods of manufacture have the boundaries set to a 95.4% confidence range and two standard deviations (2σ).
Dates provided by the analysis of monochrome black coating layers obtained from the cabinets' interior faces, collected from one or more of the drawers, are detailed in Table 3. Data suggests they are almost the same age, but probably not exactly.These results partly support the previous dating based on stylistic features but narrow the possible dates.Sko3088 appears to be slightly older than NM1, which in turn appears to be slightly older than NM26.These results partly support previous datings, based on stylistic features, but narrows the actual possible dates.The Edo era, 1603-1867, can be subdivided into shorter periods, the earliest of which coincides with manufacturing the three cabinets.
For valid dating, undisrupted and uncontaminated specimens are required.Old artefacts are inevitably used, handled, and manipulated infinite times.In surplus, sampling and specimen preparation can add contaminants.In this case, it is considered a minor problem since radiocarbon dating results somewhat agree with dating based on stylistic features.On consideration, all the specimens were sampled from unrestored interior surfaces, i.e. areas less exposed to the deposition of dirt, grime, fingerprints, and secondary material.They were handled with great care and clean tools during sampling, transport, and preparation.
Since it is unlikely that the cabinet Sko3088 was made already during the early fifteenth or the sixteenth century, data in the range of 1613-1643 more likely represents the actual time of manufacture.This time frame is earlier than the assumption based on the combination of different decoration techniques referred to above (Otawa et al. 2016) and perhaps older than the estimated date in the museum catalogue (either the 1640s or mid-seventeenth century).
Based on the combination of results, it is suggested that the cabinet dates from the 1620s to 1630s.
Evaluating the result for cabinet NM1 using the same approach leads to the likelihood that it was made during the 1630s to 1640s, narrowing the actual period for its manufacture compared with the general suggestion of the seventeenth century recorded in the museum catalogue.The scientific data does not reject the latter but narrows the most likely timeframe.
Cabinet NM26 has three different entries for manufacture in the records: 1600, the seventeenth century, and 1700-1740.The carbon dating supports the second assumption, again, it seems possible to narrow that to 1640 to the early 1660s.At the same time, the result rejects the two other assumptions.

Conclusions
The study reported used inductive reasoning, which logic has its merits but also its flaws.Any conclusion will only be valid if the studied cases are congruent with all previous results.If the studied cases are exceptions that deviate from previous observations, inductive reasoning cannot predict their features.The opposite would be to study all existing artefacts and deduce the features from the known group of cases.
In practice, such an approach is impossible.The results in this present study should be regarded as preliminary.In future, they may not hold up to scrutiny.
Moreover, limited tests do not explain or describe a cultural property's entirety, despite the technology's advanced nature.On the other hand, closely examining specific artefacts, like these three cabinets, provides a better understanding of them, even if general conclusions about all similar artefacts cannot be deduced from the results.Besides, since the artefacts are valued historical relics and cultural property, the material for analyses and position of samples were restricted.Deteriorated and peeling coatings were obtained and thus, at the same time, limited the quality and quantity of material.In the future, continued research on other namban ware in Sweden may help solve some of the issues that this study did not.
Anacard coatings are usually laminated structures with thin layers.In future, the development of improved sampling and sample preparation methods should be a subject for further research anticipated to produce even better information.Nevertheless, the present study is far more comprehensive than  most standard preparatory investigations conducted primarily concerning restoration, conservation, publication, and preparation for display, at least in a Swedish context.This study, with its imperfections, contributes new details on export urushi ware and further enriches scholarship on export urushi ware.The mixtures of saps harvested in different areas and the various blends on different surfaces on the same object indicate a diversified trade of raw material and a high level of craft skills.Indications of deliberate use of different blends of sap on the same artefact may challenge the idea that export ware was typically of lower quality and produced with less expensive materials than goods aimed for domestic purchase.This idea may still hold and warrant further scrutiny, but the past is probably more complicated than assumed.Complaints about inferior products occasionally appear in historical documents and support this supposition.However, even if that was the rule, there were exceptions.Furthermore, the threshold between the two categories may not be so sharp.Whether saps from overseas were more expensive than domestic sap was outside the scope of this study, but it is an exciting issue.Moreover, since there are indications that the deterioration rate of saps may depend on their colour and, indirectly, the botanical origin, their presence in coatings becomes crucial information for preservation.A possible future research topic may include both the frequency of sap blends used in domestic production and that for export and the long-term stability of such blends from a conservation-preservation perspective.Issues of interest for future studies also include stratified sampling to investigate the distribution of substances in different layers of ground, coating, and decoration in detail.Most routine conservation and restoration measures are carried out without comprehensive preparatory analyses in the daily management of collections.Although this will always be the case, it may be essential to consider the potential complexity of coatings.The coatings may not be what they appear at first glance.
The study demonstrates how archaeometry can contribute valuable data that can be used in many applications, not the least in re-using existing public collections.Technical investigations can prove or disprove suppositions, shed light on historical matters, and give glimpses into otherwise inaccessible cultural situations.With archaeometry, contexts can be developed for artefacts that lack such information.The three cabinets are related with very few textual sources of information, and none can tell about their early history: how they were made, where, by and for whom.Thus, to supply culture-historical contexts, other materials and methods are needed.To compose new stories that are richer, more enticing, and more accurate and to contextualise artefacts in public custody is one of the many ways to (re-)use and develop collections.These efforts may also be viewed as a form of sustainable management, using already existing resources and heritage, with the potential to gain more diverse audiences and communicate better with society.To create a socially sustainable form of collection management, custodians should not merely invite the public to 'consume' cultural heritage.When possible, they should also attempt to create a more active partnership with them to engage with collections.The more collections can be appreciated comprehensively, the better the opportunity for engaging the public.These three cabinets are prime examples of precious, finite, and nonrenewable cultural resources, which should be communicated to present and future audiences as fully as possible.
To summarise our findings, the three cabinets resemble each other because they are all made of the same or similar materials and compounds.They are Japanese products interpreted from the wood species in their bodies: asunaro and hinoki, which are endemic to Japan.
The similarities in their composition of coating structures and the modus operandi reflect traditions for urushi manufacture in Japan, including the ground layers, which are mixtures of coarse fillers of natural minerals, ceramic powders, or both.The light colour of the grounds suggests that the proportion of urushi is low, perhaps due to an admixture of a paste.The middle and top layers are transparent urushi, while metal powders were sometimes applied in-between applications of urushi or as a final application.Sko3088 appears more straightforward, with fewer layers than both NM1 and NM26, whereas NM1 shows the most elaborate strata and many layers of different compositions and functions.The density of the gold foil, powders, and flakes in assorted sizes used in a variety of techniques observed on NM1 illustrate a high level of quality in the fabrication of export ware historically, despite the general assumption that export ware was of lower quality than corresponding artefacts intended for the domestic market.
The discovery of the simultaneous use of different mixtures of saps used on NM1 exterior faces is fascinating.Perhaps, it illustrates not a limitation on costs or a shortage of local supplies but rather a deliberate and thoughtful use of saps with different physical properties.As interpreted from strontium isotope measurements, the saps in the blend used for NM1 were harvested in continental Asia.Thus, they must have been imported into Japan.The same applies to the single sap-type used on NM26.The imported raw materials may even indicate their potentially high value at the time of manufacture.Gold was encountered two times, both on NM1 and Sko3088.The gold partly lends the Sko3088 cabinet a brown hue.
In efforts to determine the age of artefacts, sometimes provenance and the earliest or first owner can be traced in inventories of assets.However, in the case of the three cabinets, such information is believed to be very laborious or even impossible to obtain.Potential records and documents that could inform on commissions or ownership remain untraced.So, despite the uncertainty and relatively wide range of the carbon dating method, that method may still be the best option available.The conclusion is that these cabinets are all products of the early Edo period.

Figure 4 .
Figure 4. Micro and macro views of the woods in the three cabinets.Above left and middle: End-grain (X), radial (R), and tangential sections (T) of NM1-1 analysed as asunaro (Thujosis dolabrata BUN 1249).Above and below right: Sample NM26-5 showing the anatomical features of the wood used in one of the drawers.The wood is assumed to be Thujosis dolabrata BUN1249 commonly called asunaro (Jap.) or Japanese cypress.The plane is almost radial with the vessels in the vertical and the rays in the horizontal.The view captures a little more than one annual period of growth.Macro view of the wood used for the body in NM26, as observed in the recess of the right drawer in the lowest tier.It has a typical conifer wood structure at a gross level, with easily recognisable principal anatomical features showing even through the black stain.Below left and middle: End-grain (X), radial (R), and tangential section (T) of Sko3088-1 tested to be Chamaecyparis obtusa B1313, native to Japan (and Taiwan), commonly called hinoki.Note the different scales.

Figure 5 .
Figure5.The edge of a wide split in the coating on the lefthand door's exterior face on NM26, displaying the coating structure.The cleavage is perpendicular to the surface, and the cupping is parallel to the wooden substrate.The edge of the detached coating displays the different upper layers in the coating system and beneath some fibres are visible on a beige or beige-reddish urushi ground.

Figure 6 .
Figure 6.Above: Thin-section of NM1-2 in VIS reflected and transmitted (left) polarised reflected and transmitted light (right) showing four-layer coating structure above wood from the body (e) with typical conifer morphology.Below: Magnified areas from the same section marked with white rectangles.Possibly cross-sections of flat fibres with small lumens.Polarised reflected (left) and transmitted light (right).

Figure 7 .
Figure 7. Thin-sections of three coating fragments from NM1-3 under VIS reflected (left) and transmitted light (right) showing typical urushi ground (c), and clear urushi coatings (a and a1).A three-to four-layer structure was observed under moderate magnification, depending on the exact position of the sample.The layer 'b' in fragment C is thicker than in B and D and shows a different character having a more heterogenous particle structure.Metal flakes (a2) are frequent in fragment C as component of layer 'a1', in XRF confirmed to be silver (Ag).Note that the scale varies from 50 to 100 µm.The white triangles mark the outmost surfaces.

Figure 8 .
Figure 8. Detail of the upper region of the same NM1-3 fragment C, as in Figure 7, further enlarged.

Figure 9 .
Figure 9. Thin-section of the coating from a fragment in sample NM1-2, further enlarged.VIS transmitted (above), VIS dark field (lower left), and polarised light (lower right).

Figure 10 .
Figure 10.Above: The NM26-2 sample before preparation, showing the wood and coating structure in which the ground and transparent coating are of approximately equal thickness.VIS (left) and UV (right) reflected light.Scale 500 µm.Below: Thinsection of NM26-1, captured in reflected and transmitted light, showing a fiveto six-layer coating structure (left).Note the fibreous structure below the ground containing a jinoko-type filler (e).The same in reflected and transmitted polarised light (right).Blue arrows indicate the outmost surfaces.The clear urushi-ground interfaces are marked with white arrows.

Figure 11 .
Figure 11.Above: Thin-section Sko3088-2, captured in VIS reflected (left) and transmitted (right) light shows a fourlayer structure and single fibres are visible in the lower region of the ground d (some marked with arrows).The transparent urushi layer b shows no interface between applications.Below: Thin-section Sko3088-1 captured in VIS transmitted light shows a thick layer of soot under the transparent yellowish and relatively thinner actual coating layer.The estimated layer thickness of the carbon black (soot) is 50-60 μm.This is assumed to render the cabinet a very deep black colour.

Figure 12 .
Figure 12.Above: XRF data from the NM1.Detection of iron (Fe), silica (Si), and calcium (Ca) are indications of a mineral compound (left).Detection of gold (Au) in high concentration and copper (Cu) indicates a gold alloy on lower carat (right).Below: XRFdata from MN26 showing a high content of iron (Fe) interpreted as an indication of a clay filler in the ground layers marked (c) in Figure 7.

Figure 14 .
Figure 14.Above: EIC at m/z = 108 from NM1 sample from interior face, detected the marker compound peaks C7 heptylphenol and C15 3-pentadecylphenol which confirm that the sap was derived from Toxicodendron vernicifluum trees.EIC from the samples collected from drawers in NM26 and Sko3088, show the same picture.Below: EIC at m/z = 108 from NM1 sample from exterior face, detected the marker compound peaks C7 3-heptylphenol and C15 3-pentadecylphenol which confirm that the sap was derived from Toxicodendron vernicifluum trees.The peaks C9 and C17 are pyrolysis products of laccol, markers for sap exuded from T. succedanea trees.

Figure 15 .
Figure 15.For the cabinets NM1 and NM26, enough sample material was obtained for strontium isotope ratio analysis.The results of NM1-1 and NM26-5 are plotted with black squares in the diagram.For comparison, other results of either Japanese or Chinese saps are shown in red.

Figure 16 .
Figure 16.EIC at m/z = 60 of direct Py-GC-MS of NM1.The peaks C16 and C18 indicate palmitic acid and stearic acid respectively.The result is similar to NM26 and Sko3088 in composition but the proportions are different.The 1,6-anhydro-β-D-glucopyranose (AGP) peak confirms a decomposition product of a starchy material (a polycarbohydrate).AGP indirectly indicates for example plant fibres or flour of unspecific origin.

Figure 17 .
Figure 17.Left: EIC at m/z = 67 and m/z = 154 of direct Py-GC-MS of NM26 sample.Pyrrole (upper) and hexahydropyrrole (lower) are pyrolysis products with amine structures detected in the coating.Right: EIC at m/z 67, m/z 81, and m/z 154 of direct Py-GC-MS of Sko3088.Pyrrole (upper), 3-methylpyrrole (middle), and isobutyl substituted hexahydropyrrole (lower) are pyrolysis products with amine structures detected in the coating.The relevant peaks are highlighted, and their molecular structure displayed.

Table 1 .
Type of analysis, specimen main content, and sampling location.

Table 2 .
Summary of the most prominent results of analyses.

Table 3 .
Results of carbon dating containing monochrome, unadorned, black coating only, sampled from one or more of the drawers in each cabinet; NM1-1, NM26-5, and Sko3088-4.