An Archaeological Challenge: The Conservation and Restoration of Luxury Roman Glass from Troia, Portugal

Abstract

A set of eleven luxury glass archaeological objects dated to the 3rd century (Roman period) and excavated in Troia, Portugal, were treated in order to restore their shapes and decorative features. In this paper, the different stages of the treatment are presented and discussed. The treatment phases—cleaning, fragments’ assembly, and development of support structures for the objects in need—had to be adapted and the treatment outline had to be revised along the process because of the heavily weathered glass that showed extreme fragility. The treatment highlights the internal support structures—consisting of thin and light structures built with thin canes of glass—that were developed in borosilicate glass for three of the archaeological objects, ensuring their physical stability for handling and future exhibition. The structures were then attached to the archaeological glass with Paraloid B72 adhesive. The treatment of these outstanding artefacts and, in particular, the development of these structures heavily depended on interdisciplinary and teamwork that resulted in the completion of the treatment for all objects ensuring their future exhibition.

1. Introduction

1.1. Troia Site and Glass Set

Located on the southwestern coast of Portugal, just south of Lisbon, the archaeological site of Troia is situated on the left bank of the Sado River estuary. The site occupies a 20-kilometre-long sandbank that separates the estuary from the Atlantic Ocean (Figure 1). Known primarily for its extensive fish-salting production facilities and the manufacturing of fish sauces, Troia remains only partially explored. As far as the authors know, Troia is the largest known fish-salting production centre in the Roman Empire, and it played a pivotal role in maritime trade and the socioeconomic landscape of the time, employing an estimated over 1000 workers. Its strategic geographical location made it a key hub within the Roman Empire. Ongoing excavations have revealed a variety of structures that attest to continuous habitation from the first half of the 1st century CE to the 6th century CE. In addition to its renowned fish-salting workshops, Troia contains evidence of a vibrant settlement. Discoveries include residential buildings, a thermal complex, hydraulic installations, a Paleo-Christian church, and several necropolises. Archaeological studies have identified various types of tombs within these necropolises, including the mensae tombs—a distinctive type of Roman burial featuring table-like structures, often semicircular or rectangular, designed for funerary banquets and offerings—which are relatively rare in the Iberian Peninsula.

In September 2021, archaeological excavations at the Roman fish-salting centre in Tróia, located across the Sado River from Setúbal (Portugal), uncovered an exceptional elite funerary glass assemblage dated to the 2nd/3rd centuries CE. This unique discovery included 11 glass objects of remarkable quality, featuring rare decorative features, and unprecedented shapes. The tomb, identified as Tomb 20, is an atypical yet carefully constructed structure, with an abundance and arrangement of grave goods that stand out. Interestingly, the absence of ceramics and the chosen funeral rite of cremation further highlight the exceptional nature of this burial. Additionally, the identity of the deceased—a woman named Flavia Matrondiae, aged 99—offers rare and valuable insights into her individuality, but also into Roman burial rituals yet to be explored. The 11 objects are briefly described in

Table 1. Photograph before the conservation treatment, accession number, and brief description of the object’s shape and decorative features. The given objects’ dimensions refer to the after-treatment dimensions. The given width is always the largest one. © VICARTE, Photos by João Krull.

Photos of the Objects Before Intervention	Brief Description of the Typology
	Kantharos with inverted conical cup and a foot. Two applied chain handles. The foot has an applied trail (very thin), apparently using the same glass as the one used for the object. No pontil mark visible (probably polished). Blown glass, very thin, completely transparent and colourless.
Accession No. 21675 [10 cm height × 13 cm width]
	Kantharos made with colourless glass. Bowl with two swing handles on stem with hollow conical foot. The cup has an out-turned thin lip, with a thin vertical wall and rounded bottom with an applied single concentric trail. The two swing handles are placed on opposite sides of the body and go from the top of the rim to the bottom of the bowl. The top of the handles is decorated with three superimposed threads of different sizes worked almost as festoons. The stem is short, solid, and cylindrical, and it has an applied hollow conical foot with a single concentric trail on the upper surface and round edge and no pontil mark visible (possibly polished).
Accession No. 21676 [9 cm height × 9 cm width]
	Flask with a blown rounded rim; neck flaring to rim and to an inverted conical body. The object seems to be undecorated. The foot is attached to the object with a round solid knop.
Accession No. 21677 [19.5 cm height × 9 cm width]
	Tall beaker with long cylindrical body that tapers to a rounded bottom with a solid stem shaped as a sphere and a foot with pontil mark. Body is made in purple coloured glass (and the foot is on colourless transparent glass). The body of the beaker is gilded with a geometric pattern of lozenges with squares in the middle (resembles a geometric pattern from architecture). On the top, close to the rim, it has a Latin inscription VTERE. FELICITER. (Use Happily)1. It has a thin string in opaque white glass applied below the rim. Below the rim it is possible to perceive an inscription.
Accession No. 21678 [15.7 cm height × 7 cm width]
	Small beaker or cup, blown and made with colourless glass. It has a pear-shaped body with a round lip. The base is almost flat with a concentric or circular applied ring foot with the pontil mark in the centre.
Accession No. 21679 [5.5 cm height × 6.7 cm width]
	Set of fragments of colourless glass, initially thought to be part of one of the other more complete objects. After conservation, it revealed to be a small beaker or cup with a pear-shaped body and a narrow neck. The body is wider at the base and the upper part tapers before flaring out to form a rim with a thickened and rounded rim. It has a concave base where an applied ring foot is visible. At the centre a pontil mark can be perceived. The decoration was made by applying two colourless threads, one below the rim and the other close to the base of the object.
Accession No. 21680 [7.3 cm height × 6 cm width]
	Bucket-shaped bowl with swing handle. The rim is rounded and out curved. The body has a biconical shape and the base is composed by a folded ring. Two circular suspension ring handles are applied to the rim to hold the handle that is shaped like an arch and is twisted. It is simply decorated with two applied glass threads, one close to the rim and the other close to the base.
Accession No. 21682 [16 cm height × 13.5 cm width]
	Flask with a cylindrical shape made with colourless transparent glass. The body has four vertical deep indentations of unequal length and depth, and a very thin applied trail below the rim is the only decoration. The base is made by a foot with a truncated-conical shape and has a discrete pontil mark in the centre.
Accession No. 21683 [14.5 cm height × 4 cm width]
	Smal beaker or cup made with blown colourless glass. It has a pear-shaped body with a round lip. The base is almost flat with a concentric or circular applied ring foot with the pontil mark in the centre.
Accession No. 21684 [6 cm height × 5.5 cm width]
	Set of fragments of colourless glass, initially thought to be part of one of the other more complete objects. Small footless beaker or cup made with blown colourless glass. It has a wide truncated conical body that surpasses its hight. The object has four deep indentations as decoration, plus a thin glass threat just below the rim. The base is flat, and the rim seems to be cut and left unpolished.
Accession No. 21685 [6.5 cm height × 8.5 cm width]
	Flask made with a colourless glass has a body with a spindle shape, supported by a baluster-shaped stem and a discoid foot with a pontil mark at centre. The neck is thin and elongated with a cylindrical shape and flares out to a round rim. The decoration of the object is made by an intricate application of colourless (base and rim) and coloured (body) glass threads. The central part of the body is decorated with yellow, red, and blue threads following a floral motif.
Accession No. 21686 [25 cm height × 7 cm width]

. The tomb (Figure 2) has a rectangular structure, excavated on the floor.

1.2. Challenges of Treating Archaeological Glass

The conservation of archaeological glass has long posed technical and ethical challenges due to its intrinsic fragility, chemical instability, and historical significance. Roman glass is frequently found in burial contexts, where environmental factors such as soil composition and humidity strongly affect its preservation state. Over the last few decades, the field has evolved from prioritising visual preservation to valuing the retention of original material, especially surface-altered layers that hold critical archaeological information.

This paper focuses on the conservation treatment of a set of 11 Roman glass objects recovered from Tomb 20 at the necropolis of Tróia Peninsula. These objects, listed and briefly described in Table 1, include items such as a handled bucket (accession no. 21682), bowls, and unguentaria. All were buried for nearly 2000 years in a dynamic coastal environment marked by saline water, sandy soil, and periodic wetting and drying cycles. These conditions caused severe glass weathering, with some objects exhibiting full-thickness alteration and extreme mechanical fragility.

Given their advanced state of degradation, these objects exemplify the most pressing conservation issues in archaeological glass today: flaking and detachment of surface weathering layers, fragmentation, structural instability, and the risk of further loss during handling or treatment. These challenges require a nuanced approach that balances stabilisation, readability, and minimal intervention.

Intrinsically fragile, archaeological glass suffers primarily from physical forces and chemical degradation. Mechanical stress—caused by pressure, impact, or soil movement—typically results in breakage and loss of fragments. Although an extensive review of glass weathering studies is out of the scope of this paper, it is important to share some of the latest research in the field because the conservation state of the weathered glass will ultimately impact the conservation decision-making process. Chemically, glass undergoes ‘weathering’ due to prolonged exposure to moisture. This process involves the exchange of alkali cations (Na⁺, K⁺) in the glass with protonic species (H⁺, H₃O⁺) in water, leading to the formation of an alkali-depleted, silica-rich surface layer. This altered layer is often iridescent, cracked, or cloudy, and while it reflects deterioration, it also preserves vital information such as tool marks and decorative traces [1].

In the past, treatments for archaeological glass often involved removing weathered layers to restore the objects’ original appearance. However, current research has demonstrated that removing these layers entails the loss of the glass’s most original and historically significant material. This loss is further compounded by the fact that these alkali-depleted weathered layers serve as protective barriers for the underlying, non-degraded glass [2], p. 17. As a result, the removal of weathered layers is no longer recommended. On the contrary, when these layers begin to detach, conservation professionals should focus on consolidating them to preserve their connection to the glass substrate and ensure the long-term stability of the object [3,4].

From the eleven objects, only one was intact with no breaking lines or detached fragments (accession no. 2169). After the objects’ removal from the tomb, the set was immediately transported to the laboratory of conservation and restoration of glass and ceramics at the Department of Conservation and Restoration (DCR) at NOVA School of Science and Technology (NOVA FCT). The laboratory is also part of the Research and Development Unit VICARTE—‘Glass and Ceramics for the Arts’.

It is usually then, when treating archaeological glass, that flaking surfaces that also present attached soil, loose fragments, and breaking lines, and gaps or missing parts, are dealt with. The treatment phases are often as follows: surface cleaning and consolidation of the flaking weathering layers, attaching fragments, and loss compensation to ensure physical stability of the object [2]. Finally, a storage and transporting container may be necessary as well. All phases should respect the principles of a minimal intervention, retractability or reversibility, and reconstructions of missing parts should be identifiable from a close distance to avoid misinterpretations with the original object, providing a successful reading of the object at the same time [3].

Regarding the materials used in the different stages of glass conservation, these will be further explored and discussed in relation to the case studies in Section 2 below. To give a general perspective, currently, it is widely accepted that acrylic consolidant offers the most suitable properties for treating archaeological glass. Specifically, the acrylic adhesive Paraloid^® B72 (PB72^®) is considered the best polymer for various treatment stages. It is used for consolidation and bonding fragments, allowing conservators to control its composition and adjust its properties. PB72^® is highly stable, reversible, and maintains its characteristics over time. More recently, it has also been used for gap filling.

Over the last 20 years, epoxy resins have fallen out of use in glass conservation, due to their excessive strength—far greater than that of glass—leading to internal stresses, especially in fragile archaeological glass. Additionally, they undergo photodegradation and yellowing due to continuous polymer crosslinking, which increases their strength and discoloration over time [3,5,6,7,8]. Finally, their toxicity is a major concern, as they are typically composed of bisphenol A, a known endocrine disruptor and carcinogen [9,10].

Finally, the glass from these objects has yet to be characterised to determine its chemical composition. Additionally, the burial soil or sand will be further analysed to measure its pH and electrical conductivity, providing an estimate of the environmental salinity. These analyses aim to correlate the intrinsic properties of the glass with the extrinsic environmental conditions, offering a deeper understanding of the weathering processes these glasses have undergone.

The purpose of this paper is to share the conservation treatment to the set of 11 archaeological glass objects in a severe state of weathering. The phases of cleaning and consolidating, and assembly of fragments, will be described. Finally, a solution of 3D support structures made of borosilicate glass was created to provide structural stabilisation for the objects in need, with the aim of respecting their archaeological nature, respecting the principle of minimal intervention, and exploring more sustainable conservation practices.

2. Treatment Preparation and Working Phases

2.1. Training Workshop

Stephen Koob, Chief Conservator Emeritus of The Corning Museum of Glass, was invited by DCR and VICARTE at NOVA to assist with the conservation and restoration of the excavated glass from Troia. A two-week workshop was planned to focus specifically on the use of Paraloid B-72, as mentioned above, for the reassembly of the glass vessels (Figure 3). In December 2021, two full days of lectures and demonstrations were presented to five post-graduate conservation and restoration students with master and PhD theses focused on glass conservation (four PhD students and one master’s student), invited by DCR. This preliminary introduction was followed by 2 full days of hands-on practice assembly of broken modern very thin glasses. The primary objective of this training workshop was to familiarise the participants with using, applying, and manipulating Paraloid B-72 adhesive before working on the actual glass objects.

Once the workshop was completed, the participants were assigned one of the Troia archaeological glasses. The glasses were then documented and treated individually, starting with cleaning, the removal of surrounding soil and debris, sorting the broken fragments, and cleaning the edges prior to reassembly (described above). The two supervisors, Stephen Koob and Inês Coutinho, were always present and available for assistance.

2.2. Cleaning

Surface cleaning and consolidation of the flaking weathering layers are two different operations that sometimes must occur simultaneously. When it comes to cleaning, the first crucial step is to determine what qualifies as dirt—what needs to be removed—and what external material should be preserved. Although it seems to be a straightforward step, cleaning is probably the most irreversible step in conservation—no one will ever be able to put any dirt removed back in the same spot! For this reason, it is of paramount importance to understand which material may be removed from an object, ensuring that it will not compromise the object’s stability and its further conservation, study, and interpretation. The basis for the action of cleaning in conservation is built on the principle that the removal of dirt will strongly contribute to the object’s stability and is normally seen as a preventive conservation measure. For instance, removing dust prevents humidity from accumulating in those particles, creating a perfect environment for microorganisms to proliferate [11], p. 96. Focusing on archaeological artefacts, the aim of cleaning can then be summarised as an attempt to remove exogenous layers of materials to achieve the original surface of the object. However, much information may be retained in between the layers of soil and the weathering layers in the case of archaeological glass. As Caple and Williams point out, it is up to the conservator’s judgement to decide on the extent of the cleaning and removal of exogenous material [11], p. 96. This judgement should then balance the age and archaeological context, the conservation state of the glass, and the information retained in the soil and hidden by it at the object’s surface. It is then wise and well accepted these days that archaeological objects are clean up to a point of equilibrium, where this step contributes to the chemical stabilisation of the material and an improvement in its aesthetics to allow for a more comprehensive study and fruition of the object.

As mentioned before, after being excavated, the objects were immediately moved to the conservation and restoration laboratory at DCR, NOVA FCT. The archaeologists removed the objects, maintaining a generous layer of soil still attached on the outside of the objects as well as all the soil on the inside. This careful removal, leaving the last stage of excavation to the conservator, provides the basis for a successful treatment. This method allowed for the recovery of most fragments, resulting in the effective reconstruction of the complete profile for all the 11 objects. The cleaning process involved excavation, during which adhered soil was carefully removed using at first soft-bristle brushes, fine paintbrushes moistened with distilled water, and dental tools. The removed soil was then placed in sealed plastic bags. Notably, in this case study, some soil remained intact, preserving the shape of the object and its degradation layers. As a result, these preserved soil formations were kept and may be incorporated into future exhibitions for the purpose of storytelling about the assemblage. Following on from what was basically a mechanical dry cleaning, upon observing a remaining layer of soil attached to the glass surfaces, it was decided to implement a chemical cleaning, which would then force a breaking of the bonds between the soil and the glass. The wet cleaning started with plain distilled water that proved effective to most of the adhered soil. However, in some areas the soil was so attached that water was not enough. This was the case of the bucket (assession no. 21682), where the glass was so weathered that we estimate that no pristine glass is present anymore, and the glass weathered through all its thickness This results in a rough, almost porous glass, which is very difficult to clean. Weak solutions of nitric acid of 3% and 6% were prepared and used to clean these stubborn soil areas [2], p. 43. The present stains were highly likely to be a mixture of carbonate deposits with soil. Nitric acid will react with calcium carbonate to form calcium nitrate (Ca(NO₃)₂), carbon dioxide (CO₂), and water (H₂O). The procedure was to apply this solution by brush in the areas where the soil was strongly adhered, leave it for two or three seconds, and then, using a clean brush, rinse it thoroughly with clean distilled water. The cleaning step should be performed four or five times after using the weak nitric solution to make sure all acid is removed and no remains are left in the object. Although used in a very weak solution, nitric acid could leach alumina and sodium and calcium oxides from the glass [12]. However, even after this use of the nitric acid, as seen in Figure 4, some stains were still present. After careful observation and evaluation, it was decided that the remaining stains were embedded in the weathering layers, and it was time to stop cleaning and accept that the brownish/yellowish stains were now part of the objects’ biography and any attempt to keep removing them would just damage the glass and trigger unwanted degradation mechanisms.

To sum up, the cleaning process was an iterative process where the need to evaluate every step was crucial to ensure that no overcleaning took place. As a rough estimation, the first step of dry cleaning removed 80% of the soil, and wet cleaning with plain distilled water was successful at removing an additional 10%. The weak nitric acid solution was effective in removing most stubborn stains and took the cleaning procedure up to a 98% effectiveness, where accepting the presence of some soil and knowing when to stop was crucial. In

Table 2. Treatment solutions, composition, application sequence, and number of cycles.

Step	Solution/Material	Concentration/Preparation	Application/Number of Cycles
Dry Cleaning	Mechanical tools (brushes, dental tools)	N/A	Initial step, removed ~80% soil
Wet Cleaning	Distilled water	Pure	Single or multiple applications to remove adhered soil
Chemical Cleaning	Nitric acid	3% and 6%	Applied with brush, left for 2–3 s, rinsed with distilled water, repeated 4–5 times
Consolidation	Paraloid® B-72	5–10% in acetone	Applied under magnification to flaking areas. Each flaking area was consolidated once, and its effectiveness was visually inspected after. Whenever necessary, a second application was performed.
Adhesive	Paraloid® B-72 + fumed silica	~50% in acetone	Applied to break edges using aluminium tubes.
Join Reinforcement	Japanese paper + PB-72	10% PB-72 in acetone	Applied internally on 3 vessels (21680, 21682, 21686)

, all for solutions used for the treatment, including for consolidants and adhesives, are summarised.

2.3. Consolidation and Assembly of Fragments

Consolidation is an invasive and irreversible treatment, applied only when necessary, typically in archaeological glass with unstable, flaking weathering layers [2], p. 55 [3]. Since glass is nonporous, the consolidant cannot be fully removed once applied; hence, the selected material must be chemically stable and well documented in conservation practice.

For the Troia glass assemblage, consolidation was essential in multiple cases to stabilise the lifting surface layers. Paraloid^® B-72 (PB72), an ethyl methacrylate–methyl acrylate copolymer, was used due to its established performance as both a consolidant and adhesive in glass conservation [2,3,13,14,15,16]. Its moderate strength, optical clarity, light stability, and solubility in acetone make it particularly suitable for fragile archaeological glass. Briefly, it has a glass transition temperature (Tg °C) slightly above 40 °C, moderate strength (although the preparation, additives, and application procedure can impact the polymer strength), and over the years it has proven to be light-stable, meaning that it keeps most of its properties without yellowing after years of light exposure [13,16]. Paraloid^® B-72 is reversible; however, consolidation is an irreversible step, independently of the chosen consolidant.

Consolidation was conducted under magnification to reduce soil entrapment between layers. Each flake was carefully cleaned on one edge before the consolidant was introduced, allowing the remainder of the layer to be treated after partial stabilisation. While complete soil removal was not always possible, the intervention effectively prevented further detachment, with no subsequent failures observed.

A particularly complex case involved object 21678, a gilded glass vessel. During excavation, a PB-72-impregnated gauze was applied by the archaeological team to secure detaching gilded layers. Removal required placing the object gauze-side-up and gradually applying acetone to dissolve the adhesive, ensuring minimal disturbance to the decorative surface. The underlying flaking layer was then successfully consolidated, restoring the legibility of the gilded design.

2.3.1. Preparing and Using Paraloid B-72 as an Adhesive

The assembly of glass fragments—particularly thin and heavily weathered ones—requires a stable, reversible adhesive. PB72 is widely recognised as the adhesive of choice for such applications due to its chemical and mechanical properties [2,3,13,14,17]. It can be formulated in a range of viscosities to suit various joining tasks and is compatible with ethanol and acetone, allowing for the adjustment of setting times and easy clean-up. PB72’s removability with acetone also permits re-alignment of joins or full reversal if necessary. Precise application was achieved using fine-nozzle aluminium tubes, which allow for the controlled deposition of adhesive along fragile edges. These tubes are filled with freshly prepared PB72 solutions and sealed for long-term use.

The adhesive was prepared in-house using PB72 pellets dissolved in acetone to a concentration of approximately 50% (w/v). To improve handling and reduce issues such as stringing and bubbling, fumed colloidal silica was added to the mixture [17,18]. The solution was stored in plain aluminium tubes and sealed by crimping, which enabled prolonged storage and facilitated controlled application during treatment.

2.3.2. Adhesive Application Strategy

Fragment joining was conducted using a “piece-by-piece” approach. Due to the advanced weathering of the Troia glasses, preliminary taping or dry fitting was not feasible. However, as many fragments had been “block lifted” during excavation, their relative positions were partially preserved, aiding reassembly.

The adhesive was applied to one break edge, and the fragments were joined and positioned on a stable surface. Initial excess was cleaned from visible surfaces immediately using acetone. In cases of slight misalignment, joints were reactivated and repositioned by locally applying acetone with a brush. Where necessary, fragments were separated and rejoined. Mechanical adjustment methods (e.g., heat) were avoided to prevent damage to the fragile surfaces.

For three vessels (accession numbers 21680, 21682, and 21686), reinforcement of the joins was necessary due to the extreme thinness of the glass, an original feature of their manufacture, further compromised by severe weathering. Thin strips of Japanese paper, embedded in 10% PB72 in acetone, were applied perpendicularly to the joins on the interior surfaces, remaining invisible to the viewer.

During assembly, it became evident that the complete profiles of all objects could be reconstructed, an essential step for future studies on provenance, production, and function. However, in some cases, material losses, particularly in the vessel bodies, prevented full reconstruction, requiring further decisions on how to address these gaps. The next step would then be to decide how to overcome the material losses.

2.4. Building 3D Support Structures in Borosilicate Glass Lampworking Technique

As previously noted, several objects presented significant structural gaps that compromised their physical stability. Given the extreme thinness and severe weathering of the glass, traditional gap-filling methods were deemed unsuitable. Instead, a non-invasive approach was adopted: the creation of internal support structures.

Support structures are increasingly common in the conservation of glass and ceramics, particularly with the advancement of 3D printing technologies [19,20,21,22]. However, each case demands a tailored solution that considers factors such as the object’s fragility, the long-term stability of the chosen materials, cost, technical feasibility, and exhibition requirements.

For the Troia assemblage, borosilicate glass was selected as the support material due to its mechanical strength, thermal and chemical stability, and optical compatibility with archaeological glass. This method, developed collaboratively by the Laboratory of Conservation and Restoration of Glass and Ceramics and VICARTE R&DU, was made possible thanks to the facilities and expertise of technical glassworker José Luís Liberato.

The process began with the design of a support structure, based on the weight and fragility of the fragments and the desired contact points. In some cases, such as object 21680, dental wax moulds were used to visualise the object’s symmetry and guide the design. Drafts were then refined through collaboration between the conservator and the lampworking specialist.

Using the lampworking technique, borosilicate prototypes were fabricated and iteratively tested and adjusted until an optimal fit was achieved (Figure 5). The final structures were adhered to the archaeological fragments using Paraloid^® B-72.

In the case of object 21682, the internal support structure had a dual function: structurally compensating for the heavy, intact handle (found inside the vessel) and providing a mount for future exhibition, reducing handling and associated risks.

While the method’s primary limitation is its dependence on specialised equipment and personnel, its advantages are significant. Borosilicate glass provides robust structural support with minimal handling during treatment, reducing the risk of damage. Unlike many infilling materials, it introduces no irreversible components; the structure is fully removable thanks to the use of B-72 adhesive. The mechanical compatibility of glass-to-glass contact—especially in terms of thermal expansion and optical consistency—was a further advantage, minimising material tensions and visual intrusions.

Chemically, borosilicate glass is among the most stable compositions available, ensuring long-term conservation compatibility. The intervention respects the authenticity of the object by visibly preserving material losses, allowing the object’s biography to remain legible. Compared to epoxy resins, this approach appears more sustainable and less hazardous to both the environment and conservators. While further data are required to evaluate long-term performance, the reversibility and physical resilience of this solution indicate strong potential for broader application in the conservation field. When evaluating alternatives to borosilicate glass supports, several strategies were considered. Three-dimensional-printed resin structures, for instance, offer precision and adaptability but raise concerns regarding long-term chemical stability, potential off-gassing, and difficulties in achieving optical neutrality. Moreover, many resins used in 3D printing are not fully reversible and often require complex post-processing steps that may introduce additional materials or residues. Reversible infill materials, such as waxes or acrylic putties, provide a lower-tech solution but frequently lack the necessary mechanical strength to support thin, weight-bearing archaeological glass without significant risk. Non-structural display aids, such as external mounts or cradle supports, were also considered; however, these do not address internal fragility and may result in visible visual interruptions or constraints in display flexibility. Ultimately, the choice of borosilicate glass was driven by its unique ability to provide internal, minimally visible, chemically inert, and physically robust support while maintaining full reversibility using Paraloid^® B-72. This decision was also guided by the conservation team’s expertise and access to specialised facilities, ensuring precision fabrication tailored to each object.

In Figure 6, the photos of the objects after treatment are presented.

3. Conclusions and Future Perspectives

This paper presents and discusses the treatment choices for a set of 11 archaeological objects dated to the Roman period and unearthed in Troia, Portugal. The artefacts are of a luxurious kind that were part of the set chosen to follow a 99-year-old woman, Flavia Matrondiae, into her grave. Due to the proximity with the Sado River—in turn, very close to the Atlantic Ocean—the archaeological glass showed severe signs of weathering that needed a special treatment design. One of the most relevant phases was the training workshop focused on archaeological glass conservation. This workshop was attended by five post-graduate students from DCR and VICARTE, who helped in starting the treatment of this set. At the conclusion of the 2-week work period, most of the Troia glasses had been preliminarily treated, and two objects were fully completed. The training of dedicated glass conservators is essential, and this was a milestone achieved in this period. The participants and Stephen Koob left, but DCR and VICARTE were able to retain Beatriz Borges (one of the trainees) to continue working on the glasses along with Inês Coutinho.

Paraloid^® B-72 proved highly effective, both as an adhesive and consolidant, allowing the fragile objects to be safely handled and assembled. Equally significant was the ethical and practical debate around cleaning and loss compensation—particularly how much to intervene and how best to respect the archaeological integrity of the weathered glass. These discussions highlight the essential role of critical thinking and continuous evaluation in conservation practice.

One of the major innovations in this intervention was the use of custom-made borosilicate glass mounts to support the most deteriorated objects. These supports provided stability without compromising visibility or historical authenticity, and they represent a promising avenue for future practice. The uniqueness and quality of the final results have secured the inclusion of the objects in the upcoming long-term exhibition at the National Archaeological Museum in Lisbon.

Looking ahead, several lines of research and development are being considered to further improve the treatment of highly weathered archaeological glass. Three-dimensional technologies (including scanning, modelling, and printing) offer enormous potential not only for designing custom mounts and replicating missing parts but also for digitally reconstructing objects in virtual environments. Further research is needed into alternative consolidants and adhesives, especially bio-based or low-toxicity materials, which can match the performance of current polymers while improving sustainability and reversibility. Studies combining material science and digital heritage could refine diagnostic tools to better assess glass degradation before intervention, especially through non-invasive imaging or spectroscopic techniques.

Finally, a comparative study across similar burial environments could help form predictive models of glass weathering under various soil and climatic conditions, informing conservation planning at earlier stages.

As this project has demonstrated, successful conservation requires technical skill but also an iterative, reflective approach. Future innovations must remain grounded in ethical principles while embracing new technologies that enhance both the preservation and accessibility of cultural heritage.