3D Digital Libraries for Maritime Archaeology: 17th- and 18th-Century Dutch Ocean-Going Merchant Ships

ABSTRACT

Digital 3D surveys of shipwrecks have become ubiquitous in recent years. As a substantial number of 3D shipwreck surveys accumulate, there remains a largely untapped opportunity for longitudinal and quantitative analyses of historic vessels in ways that were previously difficult or impossible using traditional methods, and which would provide new perspectives on the nautical archaeological record. A digital reference library was developed for Dutch ocean-going merchant vessels of the 17th and 18th century, including a substantial database of 3D scans of contemporary scale models, together with shipwreck scans, manuscripts, and other historical evidence. Important lessons for future typological digital reference libraries are highlighted, particularly the challenges of estimating scale, vessel classification and the need to publish survey data in open spatial formats.

Video Abstract

Read the transcript

Watch the video on Vimeo

© 2023 Nautical Archaeology Society.

RESUMEN

Los relevamientos digitales en 3D de pecios se han vuelto comunes en los últimos años. A medida que se acumula un número sustancial de relevamientos de pecios en 3D, aún queda en gran parte sin explotar la oportunidad de llevar a cabo análisis longitudinales y cuantitativos de embarcaciones históricas, de maneras que antes eran difíciles o imposibles usando métodos tradicionales, y que podrían proveer nuevas perspectivas sobre el registro arqueológico náutico. Se desarrolló una biblioteca digital de referencia para embarcaciones mercantes de ultramar holandesas de los siglos 17 y 18, que incluye una base de datos sustancial de escaneos en 3D de modelos a escala contemporáneos, junto con escaneos de pecios, manuscritos y otras evidencias históricas. Se destacan lecciones importantes para futuras bibliotecas digitales de referencia tipológica, particularmente los desafíos de estimar la escala, la clasificación de embarcaciones y la necesidad de publicar información de relevamiento en formatos espaciales abiertos.

摘要

近年来，沉船的数字3D勘测已经变得无处不在。随着大量3D沉船调查的积累，尚蕴藏有大量机会对历史船只进行纵向和定量分析，是以往使用传统方法是很难或不可能的做到的，这将为航海考古记录提供新的视角。我们为17至18世纪的荷兰远洋商船开发的数字查询图书馆，包括一个3D扫描当代比例模型的大数据库，沉船扫描记录、手稿以及其他历史证据。这一工作突显了未来类型化数字查询图书馆的重要经验，特别是在评估比例、船舶分类方面遇到的挑战以及以开放空间格式发表调查数据的必要性。

摘要

近年來，沉船的數字三維勘測已經變得無處不在。隨著大量三維沉船調查的積累，尚蘊藏有大量機會對歷史船只進行縱向和定量分析，是以往使用傳統方法是很難或不可能的做到的，這將為航海考古記錄提供新的視角。我們為17至18世紀的荷蘭遠洋商船開發的數字查詢圖書館，包含一個三維掃描當代比例模型的大數據庫，沉船掃描記錄、手稿以及其他歷史證據。這一工作突顯了未來類型化數字查詢圖書館的重要經驗，特別是在評估比例、船舶分類方面遇到的挑戰以及以開放空間格式發表調查數據的必要性。

المُستخلص

أصبحت أعمال المسح الرقمي ثلاثي الأبعاد لحُطام السُفن واسعة الانتشار في السنوات الأخيرة . ومع تراكم عدد كبير من أعمال المسح الرقمي ثلاثي الأبعاد لحُطام السُفن، لا تزال هناك فرصة إلى حد كبير غير مستغلة وذلك للقيام بعمل تحليلات طولية وكمية للسُفن التاريخية بطرق كانت تُعتبر في السابق مُعقدة أو مستحيلة باستخدام الأساليب التقليدية، والتي من شأنها أن توفر وجهات نظر جديدة حول السجل الأثري البحري . ومن الجدير بالذكر أنه قد تم القيام بتطوير مكتبة مرجعية رقمية للسُفن التجارية الهولندية العابرة للمُحيطات والتي تعود إلي القرنين السابع عشر والثامن عشر . تتضمن هذه المكتبة قاعدة بيانات جوهرية للمسح ثلاثي الأبعاد للنماذج المعاصرة، إلي جانب المسح بالاشعة لحُطام السُفن والمخطوطات والأدلة التاريخية الأخرى . تسلط هذه المقالة الضوء على الدروس الهامة للمكتبات المرجعية الرقمية النمطية المستقبلية، وبشكل خاص التحديات الكامنة في تقدير الحجم وتصنيف السفينة والاحتياج إلى نشر بيانات أعمال المسح في تنسيقات مكانية مفتوحة.‏

KEYWORDS:

• Post medieval
• ship structure
• Atlantic

PALABRAS CLAVE:

• Post Medieval
• Estructura Naval
• Atlántico

关键词:

• 后中世纪
• 船体结构
• 大西洋

關鍵詞:

• 後中世紀
• 船體結構
• 大西洋

الكلمات الدلالیة:

• ما بعد القرون الوسطى
• هيكل السفينة
• المحيط الأطلسي

Introduction

Reliance on 3D survey techniques has greatly increased within the discipline of maritime archaeology since ca. 2009 (McCarthy et al., 2019, p. 2). The production of large numbers of 3D surveys of shipwrecks raises a new and largely unexplored possibility of creating digital reference collections of related shipwrecks to support new types of analysis. A pilot study was undertaken to develop a digital spatial library for a defined group of vessels. The study focused on Dutch ocean-going merchant ships of the 17th and 18th centuries. This paper provides an overview and update on this work, for which only a brief interrim conference report has previously been available (McCarthy & van Duivenvoorde, 2021) and draws on data gathered during a PhD study (McCarthy, 2020).

Spatial digital survey methods are particularly useful for the analysis of historic vessels. A much-quoted observation by Muckelroy (1978, p. 10) states that, for pre-industrial societies, a ship was ‘the largest and most complex machine produced’. Ships and boats are difficult and perhaps even impossible to fully express in two dimensional formats. For this reason, the value of 3D digital analysis to maritime archaeologists was recognized even before 3D modelling was a practical reality: ‘your ship is a three-dimensional structure, so why not research it in three dimensions whenever possible … eventually most of us may be working that way’ (Steffy, 2011, p. 221). Complex analyses of the spatial evidence from maritime archaeological sites built on the data from underwater surveys, while showing great promise (e.g. Drap et al., 2013, p. 389; McCarthy et al., 2019), remains relatively limited. Single site shipwreck studies, although now mainly using 3D survey data as the basis for their site plans, are still published mainly in 2D formats, with data from geophysical surveys, manual drawings, photographs and orthomosaics represented one after another on the page in a sequential narrative. In such a format, some of the greatest potential benefits of 3D survey data for the study of historic vessels are missed. Many or all these datasets could be combined in a single spatial environment using 3D modelling software or Geographic Information Systems (GIS) to maximize the insights on the spatial configuration of shipwrecks, the original configuration of the vessel and its relationship to the land and seascape it lies within.

As 3D shipwreck surveys accumulate across the world, there is another important opportunity that has remained largely unexplored: the development of large digital spatial reference collections. These are being created in other areas of archaeology and other physical sciences. The palaeontological digital collection MorphoSource.org has been running since 2014 and currently has almost 70,000 3D scans of skeletal material, with most available for download and reuse for research. No similar projects focused on spatial data have been undertaken for historic vessels. Direct spatial comparisons between shipwrecks remains difficult as a result, discouraging studies that draw together 3D data from groups of related sites.

Longitudinal quantitative studies for nautical design by archaeologists have always been rare but are critical for the scientific credibility of the discipline (Muckelroy, 1978, p. 5). Published research on ships, with few exceptions, continues to exhibit a strongly particularist nature, emphasizing the discovery of new sites, analysis of artefacts, or survey methodologies (Flatman, 2007, p. 80). Funding is undoubtedly a contributing factor. Shipwrecks are often complex and can be expensive to investigate, keeping maritime archaeologists fully occupied with a small number of sites. Shipwrecks are also often poorly preserved, making it difficult to compare fragmentary evidence against other sources in disparate formats. For many pre-industrial ship types, the evidence base remains small, and there may be only a handful of known parallels, perhaps thousands of miles distant. As a result, typologies for many ship and boat types based on archaeology or other forms of contemporary data remain embryonic.

Gathering of a substantial digital spatial dataset for any set of historic vessels of a given type may refine or undermine the very definition of that vessel type, especially for pre-industrial and vernacular vessels. Several researchers have discussed how ship ‘types’ are defined, the pitfalls of searching for ‘ideal’ vessel types, generalistic/essentialist versus particularistic approaches, and the usefulness of biological evolution as a metaphor for ship development (Adams & Rönnby, 2013; Hocker, 1991, 2004; Maarleveld, 1995; McGrail, 1995, 1998; Schweitzer, 2017; Zwick, 2013). These important debates have unfortunately had little impact on the practice of maritime archaeology (Hocker, 2004, p. 8) with the result that many archaeological investigations of shipwrecks analyse the evidence of overall design mainly to assign the shipwreck to a pre-defined archetype. Such archetypes (e.g. Indiaman, frigate, etc.) may be derived from either a historical term contemporary to the vessel, a modern academic classification or something in between. Attempts to unpick and clarify these terms require detailed narratives of archaeological, iconographic and historical sources that show inconsistent usage among different groups, drifts in meaning over time and often direct contradictions or circular logic as historical terms become entangled with modern academic definitions.

Trying to fit all archaeological evidence from ships into archetypal categories ignores the variety found in the archaeological record of pre-industrial and vernacular vessels. Historical sources provide evidence of strong regional variations in highly traditional shipbuilding cultures, occasionally heavy reliance on hired ships, the deliberate creation of multi-purpose ships and of complex life histories of individual ships which might be refitted for different uses. One extreme example of this can be seen in the early years of Dutch merchant ships trading in the Far East, where a need for greater cargo capacity at short notice was addressed by cutting some ships in half and lengthening them (Van Duivenvoorde, 2015, p. 13). With sufficient data, ships, which can also be considered as a kind of artefact, might be placed into a continuum of changing shape over time, with each individual artefact being understood as a reflection of shifting cultural, economic, and technological factors. This can be expressed very clearly in variations in ship designs, with the decks of flutes broadening during times of war, due to the need for cannon, with a corresponding increase in crew size to man them, a change that can be detected in a changing length to width-ratio (Wegener Sleeswyk, 2003, pp. 32–33, 73–74). Rather than relying on narrative archetypal analyses, quantitative approaches such as analysis of hulls, use of cluster analysis (Castro et al., 2018), morphometrics and machine learning, might let the physical evidence speak for itself. This would assess the ship designs as part of a spectrum rather than under distinct and separate types. This type of fine-grained quantitative typological analysis is undertaken by maritime archaeologists but is usually reserved only for the artefacts from the shipwreck (e.g. Jarvis, 2023), while the vessel itself is overlooked.

Development of virtual libraries of digitized source material offers one way to develop a more quantitative and fine-grained range of source material. The digital library concept has been explored at the Texas A&M Nautical Research Lab (Monroy, 2010; Monroy et al., 2011), where a substantial digital database of digitized literature, iconography, and records of recovered archaeological material relating to for post-medieval Iberian ships have been gathered and linked. There are also some existing examples of 3D collections of shipwreck surveys for particular regions, including an ongoing project in Finland (University of Helsinki, 2019), a ‘virtual museum’ of shipwrecks created by the Bureau of Ocean Energy Management (BOEM) in the United States (BOEM, n.d.) and the Bermuda 100 Project (‘Bermuda 100’, 2017). In a similar way, collections of historic vessels held in single museums have been scanned (Cooper et al., 2018). Outside of the research sphere, hundreds or even thousands of shipwreck surveys are now being captured by local groups and avocational divers, mostly to be posted online in 3D model sharing websites without further analysis of the data. These 3D collections contain local wrecks of all types, rather than vessels of a common origin and period, and are of limited use in typological studies. Most of these raw surveys cannot be downloaded in their original complete 3D formats, and a high proportion are posted with default copyright restrictions that preclude even non-profit and academic use. In contrast, the survey and excavation of Scheurrak SO1, a Dutch flute of ca. 1600 exemplifies a professional archaeological archival approach. Here, the researchers have made their complete project archive available online through the Data Archiving and Networked Services (DANS) archaeology portal, with comprehensive digital data and descriptive metadata (Bazelmans et al., 2015).

Digital spatial libraries, particularly when used in support of digital reconstructions, can directly address issues around typological classification and variation (Figure 1). Gathering of digital data for a related group of typologically similar vessels is hampered by the fact that ships were highly mobile. The archaeological evidence for ships of common origins is often spread worldwide. As the global stock of high-quality 3D scans of shipwrecks amasses, this will become less problematic over time, but it is vital that surveys are published in an accessible way and that partnerships are undertaken with avocational groups. By undertaking a pilot study for digital spatial reference libraries for historic vessels, the challenges for more ambitious discipline-wide initiatives can be explored. Many of these challenges will be unique to this subject matter such as vessel classification, spatializing the typical range of sources for vessel designs, and harmonizing material at different scales.

Figure 1. A theoretical model of digital spatial libraries for nautical archaeology, with a distorted hypothetical model and shipwreck evidence represented as a crumpled paper ball. (Author).

Dutch Ocean-Going Merchant Ships

Ocean-going merchant ships of the 17th and 18th century Netherlands provide a suitable test group for several reasons. The definition is broad enough that there is likely to be enough spatial data to support the development of a digital spatial library, although those data are widely dispersed, across shipwrecks, museums, manuscripts, and many other sources. These ships are also highly significant to world history, playing a major role in the development of the modern global economy and trade networks, in cultural contact and conflict, in colonialization and in exploration. The Dutch Republic was small but at the height of its 17th century Golden Age (Gouden Eeuw), possessed roughly half the world’s stock of seagoing ships (Israel, 1989, p. 12). The emergence of the British Empire gradually eclipsed Dutch power through the following century, just as the Dutch had supplanted the Spanish and Portuguese at the beginning of the 17th century, with ship design playing a pivotal role in these rivalries (Scammel, 1981, p. 427).

Investigations of vessels of this type have been undertaken across the globe and have been central to the development of maritime archaeology as a discipline. There has been extensive scholarship on these vessels and on their various types. Substantial studies have focused on the jacht (Burningham, 2001, 2008; Jaeger, 2001), the flute or fluitschip (Eriksson, 2014; Eriksson & Rönnby, 2012; Hoving, 1997; Ketting, 2006; Unger, 1994; Wegener Sleeswyk, 2003), the pinas (Hoving, 2012), the warship (Bender, 2014) and the East Indiaman (Parthesius, 2010; Van Duivenvoorde, 2015). These studies have highlighted that the problems of clear vessel classification into defined types was often encountered even within formal regimes such as the Admiralty and the Verenigde Oostindische Compagnie (VOC or Dutch East India Company) (Hoving & Lemmers, 2001). Indeed, these issues seem particularly frequent in the Netherlands, and it may be that the decentralized approach of the Dutch to government and merchant administration in the 17th and 18th century and the survival of bottom-based shipbuilding techniques deterred standardization even more than in other European nations. The history of Dutch shipbuilding through this period could be considered to some extent a struggle between the increasing pressures of standardization and local and even individual practices in ship construction. These ships have resisted easy classification by even the most diligent of modern researchers:

It is almost impossible to make any hard and fast statements about seventeenth-century ship types and their names – not only because these types evolved slowly over the course of the century, but also because the meaning of the names appeared to have changed. (Hoving, 2012, p. 12)

Similar quotes may be found across the modern academic literature, in studies on the design of VOC ships (Parthesius, 2010, p. 68; Van Duivenvoorde, 2015, p. 39), on Dutch warships (Bender, 2014, p. 40) and Dutch slave ships or West Indiamen (Daalder & Spits, 2005, p. 98; Enthoven, 2013, p. 52). This difficulty is not simply due to a loss of data, as contemporary writers struggled with the same problem. The two foremost contemporary authorities on Dutch shipbuilding were Nicolaas Witsen (1641–1717) and Pieter van Dam (1621–1706). Witsen (1671, p. 187) states that ‘the families of ships are often very interbred’ while van Dam complains of the VOC rates (likely the most standardized of all Dutch merchant ships) that:

One can but read with the greatest astonishment of the alterations that from time to time, now this way, then that, were made or occurred; having been understood one way this year, and another the next year. … it being greatly to be wondered at that in the course of 100 years it has not been possible to arrive at the proper understanding of any one rate. (Bruijn et al., 1987, p. 38)

Despite this frustration among both modern specialists and contemporaries, confident definitions of Dutch ship types are frequent in more general literature. Ship types such as galjoen, retourschip, jacht, fregat, and pinas (which may or may not resemble the ships represented by their English-language homophones or contemporary Dutch terms) are described in a few words, only sometimes accompanied with disclaimer about variation. It would be wrong to suggest that these terms lack any meaning, but they are often misunderstood, and overreliance on them limits meaningful engagement with the evidence base for the many thousands of ships built through this period. In reality, only the most specialized vessels had relatively consistent designs, constrained by their duties to particular forms. An interesting example of this can be seen in large early 17th century Dutch ships, with merchantmen and naval warships remaining very similar in design and occasionally used interchangeably until the advent of line of battle tactics during the First Anglo-Dutch War (1652/1653) (Brandon, 2015, p. 84).

For Dutch ships these classification problems persist despite a wealth of primary sources, including maritime art, extensive bureaucratic records, shipbuilding manuals and charters and an extensive body of modern historical and archaeological literature. Quantitative methodologies must be developed which can engage with the full spectrum of design variation and the rich patterns of innovation and adaptation that they represent. Indeed, some of the only examples of quantitative longitudinal studies on historic vessel types by maritime archaeologists have been on Dutch vessel types of this period. These include quantitative iconographic studies on jacht design (De Winter & Burningham, 2001), a quantitative analysis of a group of fluiten in a 17th century painting (Wegener Sleeswyk, 2003) and a study including a quantitative analysis of 17th century VOC ship types in Asia based on historical records (Parthesius, 2010). These studies are quantitative in the sense that they are based on objective measurements of dimensions and design features from contemporary data, across many ships. Each of these examples are based on innovative methdologies and provided robust baseline data on vessel design, but the contribution of archaeology and other physical sources was minimal contribution due to a general lack of data.

Sourcing Spatial Data

The pilot study to develop a digital or virtual spatial library of historic vessels was initiated by the author in mid-2016. Gathering 3D scans of Dutch shipwrecks, with an emphasis on ship hulls, was the initial focus. Potential opportunities to capture new 3D underwater surveys of shipwrecks were pursued for Dutch 17th and 18th century shipwrecks in Malysia, Vietnam, England, Scotland, Sweden, and Ireland. In most cases, these either proved fruitless or the condition of the wreck sites was degraded to the point that only scans of robust elements such as guns and anchors could be made. Eventually it was possible to source enough archaeological 3D data to explore the analytical potential of the 3D digital library concept through archaeological case studies, but much work was also directed to other types of source data that could be brought into a digital spatial environment. This included shipbuilder’s manuals, iconography and historical data on ship construction and dimensions. Secondary evidence of later date was also gathered including publications and digital reconstructions, as well as scans of modern reconstructions which can be considered as a form of spatial publication. Contemporary scale models proved to be perhaps the most important source of spatial for this group of ships and provided a substantial core dataset to explore the value and challenges of the spatial library concept for maritime archaeology – for that reason the scale models will be described first.

Contemporary Ship Models

For Dutch ships of this period, there are many surviving examples of contemporary scale models, built for commemoration, decoration, illustration of shipbuilding principles and even as playthings. The evidence base for nautical archaeology is complementary to some degree, with iconographic sources illustrating the exterior of the ship above the waterline, while archaeological evidence often consists only of the lowest parts of the hull (Muckelroy, 1978, p. 217) and often only internal surfaces are exposed on the seabed. Contemporary scale models of ships, although simplified and occasionally inaccurate in some respects, typically provide a complete image of the exterior of the vessel, including both hull and rigging and ‘provide a more detailed image of reality than could even be obtained from a drawing, text, or oral account’ (Hoving, 2012, p. 235). Although the metric accuracy of contemporary scale models must be treated with caution, archaeologists have relied heavily on them in their analysis of individual shipwrecks. Van Duivenvoorde draws on the 1651 model of Prins Willim¹ for analysis of the archaeological remains of the VOC ship Batavia (Van Duivenvoorde, 2015) and on models of the 1720s for analysis of the VOC ship Zuytdorp (Van Duivenvoorde, 2012, p. 101). Martin relies heavily on the evidence of the Padmos model of 1723 for his study of the wreck of the VOC ship Adelaar in Scotland, suggesting it should be similar ‘in all significant respects’ (Martin, 2005, p. 179). Maritime archaeologists studying VOC shipwrecks of the 1740s, such as Hollandia (Cowan et al., 1975; Gawronski et al., 1992) and Amsterdam (Marsden, 1972, 1974) have also relied heavily on the 1742 Bentam models and the 1747 Merkurius model. Contemporary scale models have been extensively used as a source in the full-scale modern reconstructions of Dutch East Indiamen Batavia II, finished in 1995 and open as a tourist attraction in Lelystad (Parthesius, 1994; Vos, 1991) and Amsterdam finished in 1990 and currently moored outside the Scheepvaartmuseum.

Ship models of this type are scattered in museums across Europe, often miscategorized, and held in permanent storage with sparse catalogue entries and perhaps only a single photograph available to researchers. As a result, researchers have usually considered only a handful of well-known models, mainly those on permanent display in museums in Amsterdam. A useful starting point to build a more comprehensive database was a list made by Napier (2008) who counted 19 known or suspected VOC models, although this list proved to be flawed in many ways. In many cases, it is not possible to determine whether a particular model represents an East Indiaman, a warship, or another merchant ship such as a slaver. Scale models classified by museums as one or other type or as having a particular origin are in fact clearly miscategorized and in some cases even the contemporaneity of models is uncertain.

A revised and expanded list of ship models was created, considerably more comprehensive for contemporary models of VOC ships, and also including the majority of known 17th- and 18th-century ocean-going merchant models of other types in museums. A detailed gazetteer of ‘church models’ (ship models traditionally suspended above church naves in this period) is also available in print (van der Poel, 1987). Except for certain significant models moved into museum collections, these have been largely excluded from the study, due to their limited detail and quality, as well as their physical inaccessibility. The largest groups of models of interest were located mainly in two large collections in the Rijksmuseum and Scheepvartsmuseum in Amsterdam, including those on loan and in offsite storage facilities. Another substantial group is held by the Rotterdam Maritime Museum, with individual models in local museums in towns such as Zierikzee, Vlissingen and Dordrecht, many of which are largely unknown in the maritime archaeological literature. Outside the Netherlands, models of interest were located in the UK, Belgium, Norway, Sweden, and the U.S. Although it proved impossible to access models of interest in some locations such as Sweden and Scotland due to museum staffing issues, surveys were eventually undertaken across ten locations (Figure 2). Scanning dates were negotiated through correspondence with curators and scans were completed over three trips from Australia in 2017 and 2018. In some cases, access to museum storage and downtime allowed for capture of a handful of scans of useful comparative material such as warships of the same period and data for these examples are included in the gazetteer.

Figure 2. Map of museums where ship models were scanned, including the Netherlands, Belgium, Norway, and the United Kingdom (US model location in New Jersey not shown). (Author).

Scanning and Processing Scale Models

Scanning of the scale models proved challenging. Initial tests using photogrammetry on modern scale models were undertaken at Flinders University in Adelaide. It proved impossible to capture sufficient detail on model decks, due to the complexity of the rigging as most scanning software interprets secondary surfaces as noise in the data, deleting returns from the decks. Alternative techniques were tested, including laser scanning (using a Faro Arm with laser head) and structured light scanning (using an Artec Spider) but these were also thwarted by the complexity of the rigging. Medical CT scanning (undertaken at the Flinders Medical Centre) produced high quality geometry both inside and outside the models. This could be converted for use in 3D modelling software but lacked colour data. Nevertheless, attempts were made to apply CT scanning to museum models in the Netherlands, but this proved impractical due to the cost and risk of moving delicate ship models out of the museum. Eventually a higher quality photogrammetric workflow, utilizing wide-angle prime lenses, full frame cameras and Reality Capture photogrammetry software was developed, that proved capable of resolving the rigging and the decks behind it in most cases. This was far more flexible for working across numerous museums as the method required only a camera and tripod, although it was necessary to move models into an open and well-lit space and often to request any available lighting the museum could provide. Where time allowed, a portable endoscope was also used to examine model interiors. In some museums, glass cases could not be opened, but for many of these, photogrammetry proved capable of capturing some of these models despite the conditions. Where only one side of the ship model could be viewed, the resulting scan data was digitally mirrored along the centreline of the ship to produce a complete digital representation of the hull. The same mirroring technique was applied to half models, where model builders had modelled only one half of a ship hull for mounting on a board.

Data processing for each model was undertaken in photogrammetry software into standardized textured meshes. These meshes were imported at their true size into 3D modelling software, where their rigging was digitally removed to allow a clear view of the hull (Figure 3). The hull was chosen as the focus for this project for practical reasons, although the original meshes with rigging have also been retained in the project archive alongside the original photographs.

Figure 3. The editing of scale model scans to remove rigging. Bespoke geometrical shapes (shown in green) were created for each model to remove most of the rigging, with detailed manual editing of smaller features near the hull. (Author).

Contextualizing Scale Models

A 300-page gazetteer was produced for the 60 models of interest, including orthographic renders for all those models that could be scanned as well as standardized perspective views from the stern quarter. Entries included museum accession numbers, provenance of each model, descriptions of design, details of rigging and reconstruction where known, date of the model whether known or estimated, discussion of connections to real ships, and finally a discussion of the model’s scale. A total of 41 scale models of ships were scanned, using several hundred photographs for each model (Figure 4).

Figure 4. Contemporary scale models of Dutch ships, in chronological order spanning 200 years of nautical development, with all models digitally derigged and presented at their original scale. (Author).

Creation of the gazetteer involved extensive research into each model. For some models, entire books have been written, such as Valkenisse (Napier, 2008), Prins Willim (Ketting, 1979) and Willem Rex (Hoving, 2005), but for most of the remainder, very little research had been completed and extensive new information was generated through this study. All of these data were then brought to bear in a detailed typological analysis, supported by extensive use of 3D software to compare scale models with other spatial data, at both their original size and at the full estimated size of the ship they represented. Scanned models were grouped and regrouped by date or type such as East Indiamen, Westindiamen/slavers, flutes and whalers, warships and frigates (Figure 5). The enormous potential of 3D spatial environments was evident at this stage, allowing material at completely different scales and in different countries to be brought together into a photorealistic environment. For example, all the known models of Dutch slave ships were brought together for the first time, and three of the most significant models, those created by or for the Englishman Charles Bentam, who revolutionized VOC ship design in the mid-18th century, were brought together from two museums in the Netherlands and one in Norway.

Figure 5. A visual comparison of all scanned frigate-style scale models in the digital library, shown at original and estimated full size. (Author).

A substantial improvement in the dating of models was achieved through a synthesis of all published information for each example and through examination of their physical characteristics (Table 1). This initial review of the collection provides new data on the main typological changes through this period which are often poorly temporally constrained, such as the change from round to square tops, the change in hull shapes from ‘spindle’ shaped to box-like, the divergence in design between warships and large merchantmen, the shift from heavily ornamented high sterns to plainer and lower rounded sterns, and many other smaller changes.

Table 1. Contemporary scale models of 17th- and 18th-century ocean-going merchant ships (models in rows with dark shading were scanned).

S No.	Name	Date	Museum	Registration #nr
S1	Pinas	1627	Zierikzee, NL	0695
S2	Merchant ship	1630 (approx.)	Rotterdam, NL	M52
S3	Armed flute	1640 (approx.)	Not located, USA?
S4	Armed flute	1650 (approx.)	Amsterdam (Scheepvaartmuseum), NL	A.0211
S5	Rigged model of a 30-gun pinas	1650 (approx.)	Amsterdam (Scheepvaartmuseum), NL	A.0115(02)
S6	East Indiaman/Zealand Admiralty ship	1650 (approx.)	Zuiderzeemuseum, NL	007639
S7	N16 model Prins Willim	1651	Rijksmuseum, NL	NG-NM-11911
S8	Salamander (possibly)	1652	Rotterdam, NL	N16
S9	Armed Dutch East Indiaman	1657	Greenwich, UK	SLR0365
S10	Hohenzollern model (reconstruction)	1660 (approx.)	Rotterdam, NL	M2776
S11	Warship	1670 (approx.)	Amsterdam (Scheepvaartmuseum), NL	S.1885
S12	Saenstroom Flute	1673	Wormer, NL	None
S13	Houtpoort	1675 (approx.)	Rotterdam, NL	M211
S14	Ijsbeer/Witte Beer (possibly)	1680 (approx.)	Amsterdam (Scheepvaartmuseum), NL	2016.0007
S15	Wilhelm Prins van Oranje	1690	Amsterdam (Scheepvaartmuseum), NL	A.0149(0006)
S16	William Rex	1698	Amsterdam (Rijksmuseum), NL	NG-MC-651
S17	Den Pelicaen	1703	Antwerp, Belgium	AS.1943.009.001.A
S18	Valkenisse	1717	Boston, USA	32.183
S19	D'Bataviase Eeuw	1719	Edinburgh, UK	T.1882.28.18
S20	Padmos-Blydorp	1722	Rotterdam, NL	M201
S21	Wageningen	1723	Paris, France	13 MG 10 *
S22	Zevene Proven	1723	Chatham, UK	SLR0418
S23	Jonge Jacob	1724	Antwerp, Belgium	AS.1926.007.001
S24	Oostrust	1725 (approx.)	Antwerp, Belgium	AS.1958.078
S25	Den Ary	1725	Amsterdam (Scheepvaartmuseum), NL	A.0149(001)
S26	Van Zwijndrecht warship model	1725 (approx.)	Vlissingen, NL	7714
S27	Warship of 54 guns	1725 (approx.)	Amsterdam (Rijksmuseum), NL	NR-NM-12274
S28	Walpole model	1725 (approx.)	St Andrews, Scotland	None
S29	Principe da Beira	1725 (approx.)	Lisbon, Portugal	MM.05207
S30	Rust en Werk	1734	Rotterdam, NL	M209
S31	De Alida Galey	1735	Vlissingen, NL	M209
S32	Leef op Hoop	1740 (approx.)	Glasgow, UK	Unknown
S33	120-foot Bentam model Arendal	1742	Arendal, Norway	AAM.07170
S34	120-foot Bentam model Lelystad	1742	Lelystad, NL	NG-MC-503
S35	150-foot Bentam model	1742	Amsterdam Museum, NL	NG-MC-502
S36	Bleiswijk	1746	Dordrecht, NL
S37	D’Keulse Galy	1747	Amsterdam (Scheepvaartmuseum), NL	S.1198
S38	Merkurius	1747	Amsterdam (Rijksmuseum), NL	NG-MC-652
S39	Bossen Hooven	1750 (approx.)	Sneek, Friesland	FSM-K-053
S40	De Jonge Gerrit	1750 (approx.)	Not located, USA?	Not located
S41	D’Seeve Proovensies	1750 (approx.)	Amsterdam (Scheepvaartmuseum), NL	B.0184(01)1
S42	M1976 model	1750 (approx.)	Rotterdam, NL	M1976
S43	Witte Oliphant	1755	Amsterdam (Rijksmuseum), NL
S44	Adriaan Temmerman model	1757	Antwerp, NL
S45	D’ Gerechtigheid	1761	Amsterdam (Scheepvaartmuseum), NL	S.2397
S46	D’Elisabetgaly	1762	Vlissingen, NL	15101
S47	De Barbersteyn	1767	Antwerp, Belgium	AS.1943.009.002
S48	Beverwijk	1767	Amsterdam (Scheepvaartmuseum), NL	A.2614(01)
S49	Jonkvrouwe Catharina Cornelia	1775 (approx.)	Rotterdam, NL	M193
S50	Half Model of an East Indiaman	1775 (approx.)	Amsterdam (Rijksmuseum), NL	NG-MC-973
S51	De Dolphijn	1778	Chatham, UK	SLR0538
S52	Welvaren	1780	Sjöhistoriska museet, Sweden	O 00147
S53	Goede Hoop	1780	Amsterdam (Scheepvaartmuseum), NL	A.0125(01)
S54	Vergelijking	1780	Amsterdam (Rijksmuseum), NL	NG-MC-658
S55	De Dageraad	1780 (approx.)	Not located	Not located
S56	D’Batavier	1780 (approx.)	Amsterdam (Scheepvaartmuseum), NL	A.2781(02)
S57	Wilhelm Devierde	1789	Paris, France	11 MG 9
S58	Opsterland	1790	Amsterdam (Scheepvaartmuseum), NL	S.0924(01)
S59	De Admiraal de Ruyter	1804	Amsterdam (Rijksmuseum), NL	NG-NM-2876
S60	Wachthond	1800	Rotterdam, NL	M241

Scaling Spatial Data

A comprehensive analysis of scale is critical in combining source material, but this is far from simple for the pre-modern period. Fortunately, there are certain features of ships, and specifically of Dutch ships of this period, that can provide clues on the intended size of vessel represented in sources such as iconography and scale models. Features related to human height such as doors and ship’s wheels provide an approximate scale. More precise estimates can be made from draft marks on stems and sterns. For Dutch ships of this period there was also a strong correlation between the number of lower mainmast shrouds and the overall length of a ship, as measured by the Dutch method between the stem and stern (Hoving et al., 2015). Other important evidence includes comparison of the proportions of models against known vessels of that type, such as the prevailing charters of the VOC, who required their East Indiamen be built to one of three given sizes or charters, although they changed these several times through their existence, making the date an important variable.

A major confounding factor in assessing the scale of a data source is that systems of measurement in most pre-modern Western countries were highly localized, with unique systems in single towns. A local foot (voet) in length was often defined in the Netherlands by an iron bar in the town centre. To further complicate matters, a foot in different regions might be subdivided in 10, 11 or 12 in. Nor were these systems static, only gradually being standardized in a piecemeal way towards a national standard by the end of the 18th century. Throughout the 17th and 18th centuries the Amsterdam foot came to dominate, particularly into the 18th century (Berends, 2017). It measured 28.31 cm and was divided into 11, with many of the scale models were built to a ratio of 1:33. However, analysis of other scale models suggested many other ratios were also used, at a variety of scales between 1:12 and 1:100. To manage the necessary conversions and compare outcomes, an Access database was created and all key dimensions such as length, width and depth as digitally measured using the Dutch method (see Hoving et al., 2015). Automated formulas were written for each region to convert model measurements to full size, and to calculate key proportions of length, to width to depth. This tool proved invaluable for synthesizing source data into the digital library and something similar is likely to be a frequent requirement for future attempts to build digital reference collections for historic vessels.

Iconography

Iconography, where available, is one of the most valuable sources for tracing the history of nautical design and this is nowhere truer than for early modern Dutch ships. This period saw the flourishing of Dutch art, and the emergence of a specialist school of Dutch marine painters such as Willem van de Velde the Elder (1610–93) and his son Willem van de Velde the Younger (1633–1707). However, artists were not nautical engineers, and the less experienced of them sometimes even depicted highly anachronistic features, in some cases showing 17th century ships to illustrate 18th century mariner’s accounts, while their art might itself be of unknown date. Iconography is by its nature a two-dimensional source but the majority of the artistic evidence from this period is in a perspective format, which is much more challenging to convert into spatial formats and which may not necessarily follow the laws of perspective consistently as a photograph would. Some scholars have attempted to estimate geometrical shapes from perspective views in contemporary iconography, with varied degrees of success (De Winter & Burningham, 2001; Kamer, 2002, pp. 52–74). This has not been attempted in this project, but simpler methods such as reproducing the angle of view were frequently undertaken with the help of a suitable 3D model. Orthographic renders of shipwrecks and scale models were also made to facilitate comparison with contemporary plans and manuals. Figure 6 shows two examples of spatial comparisons made between iconographic sources and other source material.

Figure 6. Top: orthographic comparisons between scale models and plans (Scheepvaartmuseum S.0247(03) CC BY 4.0). Bottom: perspective comparison between a contemporary engraving of the wrecking of VOC ship Blijdorp at Cape Verde in 1733, and a digital reconstruction based on a scale model of the same vessel (Padmos/Blijdorp model 1722 Maritiem Museum Rotterdam M201). (Author).

At the top of Figure 6 a comparison is made between plans held in the Scheepvaartmuseum of Charles Bentam’s revolutionary new designs for the VOC in the 1740s (Scheepvaartmuseum S.0247(03)) and a scale model also thought to have been made by Bentam around the same time (Rijksmuseum NG-MC-50,249). This shows that shows that the scale model is highly accurate, even in hull shape, but also highlights subtle differences, such as the placement of masts, gunports and the curve of the beakhead. At the bottom of Figure 6 is a side-by-side comparison is shown of a painting showing the wrecking of the 1723 Dutch East Indiaman Blijdorp off the coast of Senegal in 1733 (Camstrup, 1735) and digitally recreation of the same scene using a scan of a contemporary model of the same ship dated to 1722. In this case the model is likely to be a highly accurate proxy for the ship. This is a partial reconstruction of the image, with no modification of its open gunports which would certainly be closed in a storm, nor its lack of sails. However, this method of comparison effectively highlights subtle differences in design and allows a more precise quantification of the accuracy of iconography as a source material.

Analysis, Archaeology, and Dissemination

In addition to these basic comparisons, the use of digital methods and reliable data on the scale of spatial source material facilitates a wide range of analytical and quantitative methods, as well as supporting digital dissemination and public outreach. For example, the reduction of vessel shear and lengthening of beakheads reflected can be quantified, hull profile sections can be easily captured across the entire library, with vessel lines captured using the same approach, while hydrostatic analyses can be undertaken to quantify hull performance change over time, drawing on data from both the historical and archaeological record (Figure 7).

Figure 7. A: Typological changes such as a reduction in sheer and the shortening of beakheads (galjoen) over two centuries can be quantified in these orthographic renders of contemporary Dutch East Indiaman models (scaled to full size). B: A hydrostatic analysis of the Prins Willim model undertaken in Rhino with the ORCA plugin, using a NURBS surface manually moulded for best fit with the photogrammetry scan. C: Digital extraction of sections and ship lines. D: A comparison of midships stations for all scanned scale models. (Author).

Existing longitudinal quantitative data can provide context for analysis of the digital reference library. Figure 8 shows the average tonnage of VOC ships by year and was created during the study based on raw data published in the digital version of The Dutch East India Company’s shipping between the Netherlands and Asia 1595–1795 (Bruijn et al., 1979). This visualization of the data shows how VOC ships became more standardized and larger over time, with the largest charter being used much more towards the end of the 18th century. Below this, an analysis of length to breadth ratios measured directly from the scans of scale models shows that, for scale models at least, the narrowest hulls of merchant ships is seen mainly at the start of this period, although some of the later outliers with unusually narrow hulls are seen in scale models that appear to be highly accurate in all respects, such as the Padmos/Blijdorp model.

Figure 8. Examples of longitudinal data for ships of this type, both historical in the case of the graph of average annual tonnage of ships built by the VOC through this period and derived from the digital library in the case of the quantification of length to width ratios of the scanned ship models, as measured by the Dutch method. (Author).

Tracing and 3D capture of archaeological datasets for Dutch ships of this period is necessarily a slower and more complex process than scanning of ship models. Few Dutch merchant shipwrecks of a 17th and 18th century date have been 3D scanned, and only a fraction of these have had those scans made available to researchers. The archaeological record also suffers substantial gaps. It is a sad fact that historically most located wrecks of Dutch East Indiamen have had their hulls destroyed without being recorded due to the actions of treasure hunters whose expose them to the elements in their search for valuable artefacts (Van Duivenvoorde, 2015, p. 143). Around the outset of this project, the author worked as an archaeological diver on several surveys of shipwrecks of the right type such as the Dutch East Indiaman Kennemerland (1661) (Vico-Sommer, 2016) and the Drumbeg wreck in Scotland, which is thought to be a Dutch flute of ca. 1690 (McCarthy et al., 2015). Other opportunities to work on suitable wrecks such as the Dutch East Indiaman Rooswijk (1737) with the Rijksdienst voor het Cultureel Erfgoed also arose during the study. However, none of these wreck sites retained a substantial extent of preserved or exposed hull, being either mostly buried under sediment, or having lost their hull structure to marine organisms and now consisting only of scattered non-organic artefacts. Eventually enough 3D data for hulls were gathered to develop archaeological case studies to illustrate the use of a 3D digital libraries to support maritime archaeology research. Figure 9 depicts a series of archaeological case studies, each of which has generated new insights into the wreck sites and into ship typologies (data sources and credits for Figure 9 are provided in Appendix 1). These case studies are based both on original fieldwork such as a diving expedition made to Iceland by the author in 2018 to complete an underwater photogrammetric survey of a Dutch 17th century flute ship (McCarthy & Martin, 2019) and legacy data for 3D shipwreck scans captured by other researchers shared privately with the author. However, the most important factor for the future of digital maritime libraries is undoubtedly the increasing tendency among maritime archaeologists and agencies to publish their 3D data and to allow downloading under Creative Commons licences and without charge.

Figure 9. Archaeological case studies described in the text, all made possible by the 3D digital reference collection (full details of data sources and credits provided as Appendix 1).

Only a brief description of these archaeological case studies is possible here, but even this illustrates the value of the digital library approach. In the first example (A), a laser scan of the hull of the VOC ship Batavia (built 1629) was compared with photogrammetric surveys by the author of a modern reconstruction of the ship in Lelystad, Netherlands. Together with historical detail on the unique hull dimensions of Batavia, it was possible to demonstrate the Lelystad reconstruction is a poor match in hull shape, and that a much more accurate shape can found with other sources, and further refined from historical data on Batavia’s unique dimensions, modified from the prevailing charter by the shipbuilder. Beside this (B) a newly created digital reconstruction of Batavia, with its mainmast down, is placed into its seascape context, merging bathymetric lidar, underwater photogrammetry of the wreck site and data on 17th century tides in Australia to create a metrically accurate reconstruction of the wrecking event. This example has been further developed into a separate study (McCarthy & van Duivenvoorde, in press). Next (C) is an illustration of the process of creating a digital reconstruction of the wreck of the 17th century flute Melckmeyt, by combining underwater and drone photogrammetry, scale models of a similar ship and historical data. In this example, the scale model was a close proxy for the wreck in almost all regards, but its hull shape is not accurate. Dimensions of the hull were taken from preserved charters for Melckmeyt, the hull shape from Witsen’s shipbuilding manual and other details, such as an extra deck and Danish flag were added based on historical sources. The new digital reconstruction was positioned on the seabed relative to the underwater photogrammetry survey, using aerial imagery, diver depth measurements, the alignment of the keel and the exposed mast step. As with the Batavia example, this allows for the first time an informed estimate to be made of how the wreck ship sat relative to the tide and to nearby landscape, critical elements in the wrecking event. Image D shows a collection of spatial representations of related vessels generally referred to as flutes (fluiten), bootships (bootschepen), whalers (Walvisvaarders) or Greenlanders (Groenlanders). There has been extensive discussion of the design of these ships and here for the first time a substantial collection of multibeam and photogrammetric surveys of shipwrecks has been brought together with the most important scale models of this type. These wreck surveys are mainly from the Swedish, Norwegian and Finnish Baltic, an area that is particularly important for the study of ships due to the optimal conditions for preservation of timber under water. These data were either provided directly to the author, released by its creators online under Creative Commons licences or captured directly in the case of Melckmeyt.

At the bottom of Figure 9 (E), a digital analysis of the wreck of the VOC ship Amsterdam, still preserved within the beach sands in Kent, England, where it wrecked in 1748. This was created by bringing together publicly available coastal lidar and aerial imagery, archaeological illustrations of the wreck and some contemporary models from Dutch museums which are very closely related to the wreck. This provides a more detailed understanding of the structure of this important wreck (although the models are shown here slightly higher than the wreck for the purpose of the illustration). The final image (F) shows the author’s render of work originally undertaken by Tom Cousins of Bournemouth University, who was able to use data shared by the author from the digital library to analyse the relative position of hull fragments from the Swash Channel wreck. This ship was an early 17th century merchantman thought to be Fame of Hoorn, a Dutch West Indiaman wrecked in England in 1631. Here, sections of underwater photogrammetry are aligned with the interior of a scan of a contemporary ship model of 1630 held in Rotterdam. Alignment of the fragments shows a close match with the size and sheer of the model, and with the positions of its gunports. This example highlights the value of making these data available to other researchers and is just one example of several projects and researchers that have already requested and received data gathered for the project.

These case studies show how effectively a digital library can support individual wreck investigations through a spatial workflow. In all cases, these projects were developed in the free and open-source software Blender. They also show how any relevant data that can be converted into a spatial format can be brought into the analysis, whether captured through sonar, laser, photography, or tape measure. The insights possible for each wreck site vary but this approach typically provided a more accurate understanding of the design of the ship and the initial appearance of the wreck in its landscape context. New data have been generated on the events of wrecking, demonstrating for example how much of the ship was submerged at the time of sinking, and new insights gained on site taphonomy and artefact distribution.

The virtual library also lends itself to public engagement and dissemination of heritage data. Several distinct dissemination outputs were created to explore these possibilities. These included a three-minute animated virtual dive over the Icelandic flute wreck, firstly the photogrammetric survey and then a digital reconstruction created by synthesizing historical model data with historical plans and data. This can be viewed in a VR headset or on any digital device and has now been viewed online over 65,000 times on YouTube, also forming part of a major exhibition at the Reykjavik Maritime Museum between 2018 and the end of 2022 (McCarthy & Martin, 2019). Several other animations and outreach materials including 3D prints were also created from the library during the project, but the most substantial digital dissemination output is still in the final stages of development at the time of writing. It consists of an interactive virtual museum built by the author in Unity software, requiring a VR headset and hand controllers. Inside the virtual museum, all scanned models are arranged along a row of shelves in chronological order and at their original size (ranging from less than half a metre up to over 4 m). Each model can be picked up using hand controllers and examined in high resolution. This immersive experience is set in a virtual outdoor museum in an imaginary Dutch streetscape by the sea. Users can swing their arms and release the controller trigger to throw the scans into the water, where they will grow over a few seconds to the full size of the ship they represent. It is then possible to teleport around the scene and up onto the deck of each ship, where they will find a museum exhibit unique to each scan, based on the images and data in the gazetteer (Figure 10). The intention is to make this experience available online as a free game, and to include more examples of archaeological data alongside the ship models where possible).

Figure 10. A: The initial view of the virtual library of ship scans. B: The user picks up a digitized ship model. C: The user can throw the model into the sea, where it grows to full size. D: The user can teleport up a gangway onto the deck where they will find a virtual exhibition for each wreck or ship model describing its individual details, its 3D survey and the wider context of its type. (Author).

Conclusions

The development of a digital library of early modern historic ships was undertaken as a proof of concept, to explore how the acquisition of 3D data, particularly underwater photogrammetry, might be combined and compared with other sources in a spatial environment. This pilot study has provided a wealth of new insights on the scale models and other source material relating to Dutch oceangoing merchant ships of the 17th and 18th centuries.

For shipwreck data, the main challenges will continue to be the geographical spread of data (a particularly acute problem for nautical archaeology), the relative infrequency of shipwreck surveys, and data availability. The slow accumulation of high-resolution underwater sonar and photogrammetric data will gradually overcome the first two issues, but data availability can be solved more quickly, through the adoption of open data policies and repositories. Otherwise, the publication of 3D data flattened into two-dimensional formats, of 3D data in non-reusable/non-downloadable web platforms, or under highly restrictive copyright terms will continue to hamper broad longitudinal and quantitative studies of nautical typology. For this reason, the entire digital library of model and archaeological scans captured by the author will be made available under Creative Commons licence once key project publications have been finalized.

There is enormous potential for maritime archaeologists to collaborate to develop this approach much further. The library of scanned material developed here is modest in scope, for the most part the work of a single researcher over a few years and represents only the tiniest fraction of potential source material for post-medieval sailing ships. One initial and low-cost step would be for major collections of contemporary ship models to begin programmes of scanning and releasing their entire collections digitally, from the earliest anthropological collections up to modern merchant and warships. This would require substantial commitment of time and developments of defined standards but would quickly benefit maritime archaeologists researching all periods and in all areas of the world.

Fortunately, the outlook for maritime archaeology is positive. It is likely that scans of shipwrecks and related source material will increasingly become available to researchers at scale, and this will have a profound impact on the way we understand the evidence from shipwreck sites. The 3D digital library approach provides a more meaningful use for the wealth of underwater survey data being provided by spatial sonar and photogrammetric techniques. This may help the discipline of maritime archaeology to move further towards the scientific approaches that we aspire to. Longitudinal perspectives across many wreck sites will draw the focus away from the ‘particularist’ approach. Rather than examining shipwrecks to assign them to pre-existing and often flawed nautical archetypes, we will increasingly be able to locate them within a detailed spectrum of change and innovation that reflects the ever-changing political, technological, and social context of ship and boat construction.

Permissions Statement

Material presented here is published with permission of the Rijksmuseum, the Scheepvaartmuseum and Maritiem Museum Rotterdam.

Acknowledgements

I wish to thank my PhD supervisors Wendy van Duivenvoorde and Jonathan Benjamin for outstanding mentorship and support through my studies in Adelaide on the traditional lands of the Kaurna people. Funding for fieldwork was provided by the Embassy of the Kingdom of the Netherlands in Canberra and Flinders University and for conference attendance by the MAP Fund. This project would not have been possible without the many museums and other institutions and researchers who provided data, access to material and substantial staff time, particularly from the staff of the Rijksmuseum, the Scheepvaartmuseum and Maritiem Museum Rotterdam. I also thank Liam Phillips of Flinders University and Kevin Martin of the University of Iceland for support in fieldwork. A substantial debt is owed to Ab Hoving for expert input at many stages of the project. I also thank Jon Henderson, Johan Rönnby and the peer reviewers of this paper for considered feedback on written content.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Australian Research Council: [Grant Number DE220100403]; Embassy of the Kingdom of the Netherlands, Australia: [Grant Number GCE-2017-03].

Notes

1 Scale model names are given here with the spelling used on the model and not corrected or changed to match a real ship, so Prins Willim rather than Prins Willem.

Appendix 1

Data sources and credits for Figure 9

   

	Evidence	Source
A	In yellow: Laser scan of Batavia hull 1629, with best fit to related ship scans	Courtesy of Wendy van Duivenvoorde, scanned in Fremantle 2005
	Left: Batavia II reconstruction (full size) in Lelystad, Netherlands	Photogrammetric scan by author in Lelystad, April 2018
	Centre: Prins Willem ship model 1651, Rijksmuseum NG-NM-1191110, scaled up from an estimated 1:50	Photogrammetric scan by author in Amsterdam, April 2018
	Right: N16 ship model (unknown date), Maritiem Museum Rotterdam, scaled up from an estimated 1:47	Photogrammetric scan by author in Rotterdam, April 2018
	Far right: photogrammetric scan of Batavia hull	Western Australian Museum CC-BY-4.0
B	Centre: digital reconstruction by the author (Batavia III) with mainmast down	Modelled by author (see McCarthy & van Duivenvoorde, in press)
	Base: bathymetric lidar survey of Houtman Abrolhos	Department of Transport Fugro Landmark 5 m bathymetric lidar data, captured 2016, CC 4.0
	Above lidar: underwater photogrammetry survey from 2019	Courtesy of Dave Jackson of Diving Western Australia
	Above photogrammetry (in brown): laser scan of Batavia hull 1629	Courtesy of Wendy van Duivenvoorde, scanned in Fremantle 2005
C	From left to right:
	Underwater photogrammetry survey of Melckmeyt (Melkmeid) Dutch flute	Captured by Kevin Martin and the author in two seasons of diving 2016–18
	As above, with colour ramp	As above
	Drawing of a flute ship, in elevation and stations, aligned in 3D	Nicolas Witsen (1671)
	As above, in green	As above
	17th century flute ship model ‘Houtpoort’ with rigging digitally removed	Photogrammetric scan by author in Rotterdam, April 2018
	Digital reconstruction of Melckmeyt wreck, derived from Houtpoort model with alterations	Modelled by author (see McCarthy & Martin, 2019)
D	Clockwise from top left
	Multibeam survey of ‘Ghost ship’ flute at 125 m depth, surveyed in 2010 by MMT (Marin Mätteknik) Sweden	Courtesy of Johan Rönnby (Södertörn University), Niklas Eriksson (Stockholm University) and Joakim Holmlund (MMT Sweden)
	17th century flute ship model ‘Houtpoort’ (Maritiem Museum Rotterdam M211)	Photogrammetric scan by author in Rotterdam, April 2018. Digitally derigged and mirrored along keel
	Scale model of bootschip De Goede Hoop (Scheepvaartmuseum A.0125(01).) dated to ca. 1780, scaled up from an estimated 1:48	Photogrammetric scan by author in Amsterdam, November 2018. Digitally derigged and mirrored along keel
	Scale model of a mid-17th-century flute (Scheepvaartmuseum A.0211), scaled up from an estimated 1:36	Photogrammetric scan by author in Amsterdam, November 2018. Digitally derigged and mirrored along keel
	Scale model of whaling flute Prins van Oranje 1690 (Scheepvaartmuseum A.0149(0006)), scaled up from an estimated 1:33	Photogrammetric scan by author in Amsterdam, November 2018. Digitally derigged and mirrored along keel
	Underwater photogrammetry survey of Melckmeyt (Melkmeid) Dutch flute	Captured by Kevin Martin and the author in two seasons of diving in Iceland 2016–2018
	Underwater photogrammetry survey of a 18th century flute wreck, lying at 13 m deep near the island of Jutholmen, Sweden	Surveyed by Markku Luoto of the Maritime Archaeological Society of Finland (retrieved January, 2023, from https://sketchfab.com/mas-fi) CC-BY-4.0.
	Underwater photogrammetry survey of a 17th century flute wreck, lying at 13 m deep near the island of Jutholmen, Sweden	Swedish National Maritime and Transport Museums (retrieved January, 2023, from https://sketchfab.com/maritima) CC-BY-4.0.
	Underwater photogrammetry survey of Anna Maria (built 1694, sunk 1708) flute wreck, lying at 20 m in Dalarö harbour, Sweden	Surveyed 2018 by the Swedish National Maritime and Transport Museums (retrieved January, 2023, from https://sketchfab.com/maritima) CC-BY-4.0
E	Base 1 Intertidal elevation data: Coastal and intertidal multibeam survey of the English coast, 1 m resolution, captured with Kongsberg EM3002 Dual Head	Channel Coastal Observatory TQ7708_20130801, Open Government Licence
	Base 2 Terrestrial elevation data: LIDAR Composite DTM 2020–2 m	UK Environment Agency Open Government Licence
	Base 3 Aerial imagery (10 cm), projected by author onto merged elevation data from Base 1 and 2.	Channel Coastal Observatory TQ7708_20180625, Open Government Licence
	Left: a 3D modelled reconstruction of the current appearance of the wreck of Amsterdam (1747).	Modelled by author
	Left-middle: Digitally reworked and spatialized archaeological drawings of the wreckage of Amsterdam	after Marsden (1974, p. 128)
	Right-middle (and background): 1742 contemporary scale model of an Amsterdam VOC 150-footer, the ‘Bentam model’, Rijksmuseum NG-MC-502 (currently on long-term loan to Amsterdam Museum)	Photogrammetric scan by author in Amsterdam, on 2 May 2018. This model was a sort of advert or template for ships like Amsterdam. Both this and the model below are tilted to match the heel of the wreck of Amsterdam but are shown slightly higher than the wreckage for the sake of the illustration.
	Right (and background): 1747 contemporary scale model of a Zeeland VOC 150-footer, Merkurius, Rijksmuseum NG-MC-652. Model is digitally scaled up from an estimated 1:22	Photogrammetric scan by author in Amsterdam, on 4 April 2018. This scale model post-dates Amsterdam by a few years and is from a different VOC chamber but shows a ship of the same specifications.