Introduction

Scientific bottom trawl surveys are widely employed across the world to assess the health of marine species and their ecosystems. In some regions, fisheries-independent data have been collected from these surveys since the late 19th and early twentieth centuries following concerns about the depletion of fish populations (Meek 1895; Garstang 1905; Cefas 2014). This information is highly valuable for the reconstruction of past ecosystems states, particularly for providing baselines of the status of commercial and non-target fish species (Pinnegar and Engelhard 2008). It is difficult to capture their ‘natural’ baselines given that most trawl surveys started after the onset of widespread trawling (Hunt et al. 2024), and pre-industrial fisheries have impacted fish populations for centuries (Thurstan et al. 2014). Nevertheless, historical trawl records can be contrasted with contemporary data to provide realistic benchmarks for biodiversity restoration and rebuilding fish stocks.

Studies that compare historical and contemporary trawl catches have revealed long-term changes in the diversity, abundance and size distribution of many species (e.g. Greenstreet and Hall 1996; McHugh et al. 2011; Currie et al. 2020). However, such studies are often faced with many biases because they utilise catches obtained across different surveys, trawl gears and methods of operation (Rijnsdorp et al. 1996; Currie et al. 2019). Fish of different species and sizes exhibit very different catchability, and the trawl design will have considerable influence on the quantity and composition of the catch (Jennings et al. 2001). Beam trawls have fixed geometry and are well not suited to catching fish living above the seabed (Gunderson and Ellis 1986), whereas the shape of the net mouth in otter trawls vary with depth and towing speed (West 1982). The twine material and mesh sizes used to construct a trawl net will also affect selectivity (Currie et al. 2019), especially in the rear (cod-end) portion of the net where most fish are retained (MacLennan 1992). Here, the chance for a fish of a given size to escape is largely determined by the degree of opening of the cod-end meshes (Jones 1963), but also by the physical properties and thickness of the twine material used (Tokaç et al. 2004; Sala et al. 2007).

Understanding how trawl geometry can influence selectivity is essential for standardising comparisons between catch rates in time and space. Modern scientific surveys are often very prescriptive about trawl durations, towing speeds and gear types in order to yield consistent and comparable catch rates (Jennings et al. 2001). However, studies that attempt to contrast trawl catches over longer time frames should account for developments in fishing power and technology over the past century (Engelhard 2008). This is often achieved by standardising catch rates by applying a constant fishing power or swept-area correction factors (Rijnsdorp et al. 1996). Alternatively, it is possible to replicate original fishing practices and resurvey historical sites to investigate long-term change (e.g., Lund et al. 2011; Currie et al. 2019), however these types of experiments have not been routinely performed due to the high costs associated with trawl surveys. One example was conducted by McHugh et al. (2011) in the southwest of England, where authors resurveyed historical sites in 2008–2009 using a similar otter trawl and mesh sizes to the one employed during historical surveys in 1913–1922, revealing substantial changes in the composition of an inshore fish community. In 2015, Currie et al. (2020) resurveyed three inshore sites of the Agulhas Bank in South Africa, which were originally surveyed in 1897–1904, by replicating the design and functionality of the original trawl gear and methods of operation. They rebuilt an original ‘Granton’ otter trawl based on original net plans and dimensions, historical photographs, and wider scientific literature to accurately investigate long-term change in a demersal fish community that pre-dates extensive commercial trawling (Currie et al. 2019).

In European waters, commercial trawl fisheries have undergone rapid changes in technology, effort and target species over the last two centuries (Engelhard 2008). Early trawls, comprising of a wooden beam, were towed relatively close to port by sail-powered fishing vessels until the mid-nineteenth century (Alward 1932). The beam varied in size (6–12 m) depending on the size and power of the vessel that towed it (Collins 1889; Alward 1932; March 1953), holding open a triangular bag-shaped net ~ 1.1 m above the seafloor, secured at either side by two metal ‘trawl heads’ (Alward 1932). Sailing smacks typically worked beam trawls that were ~ 6 m in length, but they were soon displaced by steam-powered fishing vessels in the 1880s, which could tow much larger beams up to 12–15 m (Edwards 1909; Alward 1932).

Despite enhanced fishing power and easier access to deeper fishing grounds, beam trawls reached their workable limit as the manageable length (and weight) of the beam prevented an expansion to the size of the net (Heape 1887; Edwards 1909). By the end of the nineteenth century, a further increase in fishing power led to the widespread use of a new type of trawl gear, the ‘Granton’ otter trawl (Kyle 1903; Engelhard 2008). On steam trawlers, beams were soon replaced by a pair of otter boards (Fulton 1901), which act as underwater kites to keep the mouth of the net open (Holdsworth 1874; Cunningham 1896). As a result, the use of beam trawls were reduced to the shallow, southern part of the North Sea, where the last sailing trawlers targeted plaice (Pleuronectes platessa), sole (Solea solea), brill (Scophthalmus rhombus), and turbot (Psetta maxima) (Engelhard 2008). Unlike unwieldy beams, otter boards could be conveniently stowed aboard the vessel and increased the spread of the net for catching a greater diversity of species (Kyle 1903). The headline of an otter trawl could rise higher off the seabed in contrast to cumbersome beam trawls, and the width of the net could expand to a much larger volume (Kyle 1903), thereby increasing the number and diversity of fish caught. By 1898, Garstang (1900) estimated that the total fishing power of one steam otter trawler was equivalent to eight sail-powered beam trawlers.

The addition of steam power to fishing vessels during this period, which increased fishing power and their ability to tow much larger nets, had subsequent effects on local fish populations (Garstang 1900; Engelhard 2008). Around the British coast, these concerns were expressed by fishers early on, resulting in the establishment of a three-mile trawl ban, which prohibited trawling and protected inshore fishing grounds (Meek 1895; Allen 1897). This prompted naturalists to subsequently set up a series of scientific trawling experiments to investigate the impact of trawling on fish species in inshore waters (McIntosh 1895; Meek 1899b). These include trawl surveys conducted by the Northumberland Sea Fisheries Committee (NSFC) and the Dove Marine Laboratory, Cullercoats, within the three-mile trawl ban on the Northumberland coast (UK) between 1892 to 1913 (Meek 1895–1913, 1899a, b). Naturalists of the NSFC collated detailed information on the species composition, size distributions and ‘food of fishes’ using a late nineteenth century wooden beam trawl in five shallow coastal bays. Availability of these early records provided a unique opportunity to resurvey the same bays more than a century later, with the motivation to investigate long-term change in an inshore fish assemblage and identify potential baselines for ecosystem recovery (see Hunt 2022; Hunt et al. 2024). To achieve this, we set out to collate and synthesise the information required to reconstruct the design and functionality of the original trawl gear employed in the NSFC trawl surveys. Specifically, we provide a breakdown of the trawl gear components, review relevant information from the literature to describe late nineteenth century beam trawls, and outline the reconstruction process using currently available materials. Finally, we conducted a series of comparative gear trials to test the efficiency of the replica beam trawl with a modern otter trawl by comparing fish catch rates and length-frequency distributions.

Methods

Gear review and reconstruction

To recreate a comparable late nineteenth century beam trawl, information on historical trawling practices were collated by reviewing relevant literature sources. Firstly, gear-specific and survey information was acquired from NSFC reports, accessed online via the Biodiversity Heritage Library (www.biodiversitylibrary.org). Survey-specific information was also attained from local newspaper articles from the British Newspaper Archive (www.britishnewspaperarchive.co.uk). Where details were lacking in NSFC reports, information on early beam trawl design plans, gear components and materials was collated from other sources, including the wider literature, photographs, newspaper articles, and direct correspondence with the fishing industry (Online Resource 1). Literature searches were conducted using a combination of key word search terms, such as early beam trawls; nineteenth century trawl gears; net plans; wooden beam; historical net plans; mesh sizes; and net materials. Historical photographs illustrating the original gear, research vessel and deck scenes were sourced and scanned from one article written by Alexander Meek in the Windsor Magazine (1899b). Additional vessel-specific information and images for estimating dimensions were obtained from the website ‘Tyne Tugs and Tug Builders’ (http://www.tynetugs.co.uk).

We attempted to distil the information that was thought to be relevant for a possible reconstruction of a late nineteenth century beam trawl. Dimensions of the original gear were not specified in NFSC reports. Instead, dimensions discernible in historical photographs were scaled and approximately estimated using Image J software (version 1.52). Firstly, the known width of the paddle tug Livingstone was used to scale and measure the depth of the vessel’s gunwale, as it was a common feature present in photographs. Secondly, the estimated depth of the gunwale was used to scale the trawl heads and beam diameter using the measuring tool in Image J. Without knowing precise dimensions of objects within the images, it was difficult to verify this procedure. However, other features (e.g., average height of a man in the 1890s) were used to support whether the vessel and gear were approximately and realistically scaled. Data recorded in NSFC reports, such as site-specific tow durations and distances, were utilised for calculation of historical trawl speeds.

For the gear reconstruction, many historical materials (e.g., hemp twine) were no longer available due to technological changes in the construction of trawl nets. Instead, we found matches by sourcing alternative materials and components that were thought to have similar functional properties (e.g., weight and density) to the original gear (see Results for details).

Gear trials

In June 2018, initial gear trials were conducted to test the behaviour and performance of the replica beam trawl. Subsequently, the gear underwent several minor modifications to improve the catch efficiency, while ensuring that the core elements of the original design was retained. In August 2018 and March 2019, we conducted a series of comparative trawls with a modern otter trawl to further compare the catchability and efficiency of the replica gear (Online Resource 3). Gear trials were conducted in the same shallow (< 25 m) coastal bays that were previously surveyed by the NSFC in 1892–1913. The August trials were conducted in five coastal bays, namely Blyth Bay, Cambois Bay, Alnmouth Bay, Druridge Bay and Skate Roads, whereas March surveys were carried out at the former two sites only. All trawls were conducted in the daytime at similar depths in the same substrate type (coarse sandy habitat), with a standard one hour haul duration on board Newcastle University’s Research Vessel The Princess Royal. The same trawl tracks were not repeatedly trawled over in a single day of surveys.

Table 1 Summary of each gear component (including material options, dimensions, and references) for the reconstruction of a late nineteenth century beam trawl based on information synthesised from Northumberland Sea Fisheries Committee (NSFC) logbooks, wider historical literature and technical expertise. Final dimensions of the replica gear are also summarised along with adjustments made to the replica gear following initial gear trials in June 2018
Full size table

The modern otter trawl net (12.8 m total length) was composed of polypropylene and had headline length of 8.2 m and ground rope length of 12.8 m. Each net panel including the cod-end consisted of an 80 mm inside mesh. The otter boards or doors were fixed between two bridles (73.1 m length) and warps released from the gallow blocks either side of the vessel’s stern. The spread of otter doors can increase with depth due to changes in trawl geometry (Ragnarsson and Steingrímsson 2003), but this is likely to be minimal because all trawls were conducted in relatively shallow depths of < 25 m. Some authors have assumed the net opening to be two-thirds the length of the headline for historical trawl gears (Rogers and Ellis 2000), while others have used net or wing-end spread as a measure of the horizontal net opening (e.g. Trenkel et al. 2004; Broadhurst et al. 2012). The door spread (distance between the two otter boards) was estimated to be 26.5 m based on the distance between the gallow box and the amount of warp out (for methods of calculation see https://www.seafish.org/responsible-sourcing/fishing-gear-database/technical-information/calculating-the-spread-of-trawl-doors/). We estimated the wing-end spread (as a measure of the horizonal net opening) to be 4.07 m, which was calculated using the door spread, total bridle length and ground rope length (following https://www.seafish.org/responsible-sourcing/fishing-gear-database/technical-information/bridle-angle-and-wing-end-spread-calculations/). The otter trawl was towed at ~ 1.29 ms−1 and beam trawl at ~ 1.03 ms−1 to imitate historical towing speeds.

After each haul, the contents of the cod-end were sorted by species, identified, and all fish were counted, and total length (cm) measured. For the main species caught by both trawl gears, catch rates were expressed per unit area (numbers of fish per 1000 m2) per hour swept of the seabed to test whether the catch efficiency of gears differed, irrespective of the differences in towing speed. The swept area was defined as the wing-end spread of the otter trawl (4.07 m) and fixed width of the beam trawl (6.7 m) multiplied by the distance trawled (based on trawl duration and their respective towing speeds). The estimated swept area per hour of trawling for the otter trawl was 18,800 m2 and 24,800 m2 for the replica beam trawl.

To compare differences in mean catch rates of the main species caught between the two trawl gears, non-parametric Wilcoxon Mann–Whitney U-tests were applied. To avoid potential confounding effects of mesh selectivity in comparisons, we only included fish > 15 cm in analyses. Length-frequency distributions of plaice and dab (Limanda limanda) from August gear trials were examined based on sufficient data availability for comparisons between the replica beam and otter trawl. Two-sample Student and Welch t-tests were used to assess differences in the mean lengths (cm) of plaice and dab, respectively. Dab length data was log transformed to ensure approximate fit to a normal distribution. Additionally, Kolmogorov–Smirnov two-sample tests were conducted on plaice and dab separately, comparing the maximum absolute differences between the cumulative length-frequency distribution of the two trawl gears. Length-frequency distributions in March gear trials were not statistically compared for these species due to insufficient sample sizes obtained by the replica gear in this period. All analyses were conducted in R version 4.0.2 (R Core Team 2020).

Results

19th century beam trawl net designs

Review

The design of trawl nets can vary according to the species targeted, type of substrate trawled and fishing power of the vessel towing the gear (Engelhard 2008). However, the various components of a trawl are proportionate to its size (Nair 1969). For beam trawls, the net configuration is determined by the prescribed length of the beam, height of the trawl heads and ground-rope length (Collins 1889). Unlike the variable net opening of an otter trawl, the horizontal mouth width of a beam trawl is fixed, and its geometry is subject to less variability as a constant width is maintained (Kuipers 1975). However, the dimensions of the different portions of the net, including the materials and thickness of the twine, and method of connecting the net panels, will ultimately affect the behaviour and shape of the trawl net (Currie et al. 2019).

In principle, a beam trawl net in the mid to late nineteenth century comprised an elongated, triangular-shaped net (Fig. 1a; Collins 1889; Edwards 1909; March 1953). Although this same basic shape was adopted by beam trawlers all around the British coast, the precise length and dimensions of the net were adapted depending on the size and power of the vessel that towed it as well as the characteristics of the seabed where the vessel was operating (Holdsworth 1874). The total length of the net was approximately twice the beam length, while the upper part or “square” made up half the size of the entire net length (Holdsworth 1874; Collins 1889; March 1953). The square was fastened to the beam in sections by rubber grommets (Fig. 1b; Collins 1889) and typically raised 0.9- 1.2 m above the seafloor in large beam trawls (Wood 1911; Alward 1932; Butcher 1980). The underneath section or “belly”, extending from one trawl head to the other, was cut away to form a sweeping semi-circle located close the ground (Collins 1889; Wood 1911). Here, the “bosom” or centre of the curve was typically set at a depth equal to the length of the beam (Holdsworth 1874). The two corresponding narrow sides of the net, termed “wings”, extended from the trawl heads to the bosom (Davis 1927). The wings were made from individual pieces of netting and inserted after the trawl was laced together (Holdsworth 1874).

Fig. 1
figure 1

Adapted from Collins (1889), which is in the public domain

Plan view of a late nineteenth century beam trawl net, illustrating a) a triangular-shaped net with an unopened cod-end and b) a detailed net plan with labelled features.

Full size image

Cod-ends were approximately one-seventh the size of the entire length and fastened together at one end by a ‘cod-line’ (Fig. 1b; Holdsworth 1874; Collins 1889), which retained and prevented the escape of fish. Like modern trawl nets, the underside of the cod-end was protected from chafing as it was dragged over the seabed with the addition of ‘rubbing pieces’ or ‘false bellies’, old pieces of netting, which were fitted in overlapping sections (Holdsworth 1874; Davis 1927). As per the original net, late nineteenth century nets were fitted with pockets laced between the belly and back, one on either side (Holdsworth 1874; Wood 1911; Davis 1927), creating a funnel-shaped passage ahead of the cod-end for fish to enter (Graham 1956). At the entrance of the cod-end, a loose piece of netting knowns as a ‘flapper’ was laced to the upper surface and functioned as a trap to prevent fish from escaping after entering the cod-end (Holdsworth 1874; March 1953). Although they do not feature in modern trawl nets, pockets and flappers continued to serve their purpose for several decades into the twentieth century (Davis 1927; Currie et al. 2019).

In reference to the original ‘22-foot beam trawl’ employed in NSFC scientific surveys, Meek (1899b) stated that ‘The large net is conical in shape. The upper part of the broad end is fastened to the beam and the lower half is weighted, usually with a thick, heavy rope… The net is provided with pockets…”. Beyond this description, there was no specific design plan of the original beam trawl documented by the NSFC. Alternatively, we compiled various net plans and descriptions that provided detailed information on the construction of late nineteenth century beam trawl nets. The most comprehensive drawings and descriptions we found from this era were by Holdsworth (1874) and Collins (1889). These plans detailed key net features and provided an account of general net parameters relative to a trawl’s size and methods of construction. To further support these descriptions, we examined various examples of different sizes of nineteenth century and early twentieth century beam trawl nets and their specific dimensions (Collins 1889; Garstang 1905; Davis 1927). It is worth noting that these refer to large commercial nets employed at various locations around the British coast; specific details on smaller beam trawls were not readily available. We also reviewed net plans of a 13.7 m 1880’s replica beam trawl used during fishing trials conducted onboard a restored Lowestoft sail trawler the SS Excelsior (Millner et al. 1997; personal communication John Wylson, The Excelsior Trust).

Reconstruction

For the replica beam trawl, we designed a net to fit a 6.7 m beam based on the aforementioned design plans and collective descriptions obtained from multiple sources. In this study, the net design was similar to the dimensions and methods of construction of trawl nets reported in the literature (based on 12 to 15 m beam trawls), but we adjusted the dimensions and number of meshes of each net panel accordingly to fit a 6.7 m beam trawl. On this basis, the total length of the replica net was downscaled to 13.3 m. The square of the net was 7.3 m, and the cod-end was scaled to be approximately one-seventh the length of the net (2.1 m). The number of meshes were determined by the height and lengths of the various net panels, twine thickness and mesh size (these are discussed in separate sections below). Because the square of the net initially ballooned following gear trials in June 2018, the number of meshes was reduced from 180 to 160.

Based on the literature, pockets were laced to the baitings and belly and separated by the flapper at the cod-end entrance. As in historical trawl nets, the flapper was connected to the upper surface of the cod-end. Detailed plans of the reconstructed net are provided in Online Resource 2.

Net materials

Review

The majority of selectivity occurs in the cod-end of a trawl net. Apart from the overall trawl design, there are a number of other gear parameters that can influence cod-end selectivity and fishing performance, including the physical properties of different twines (e.g., specific gravity, flexibility, thickness) used to construct a trawl net (Online Resource 1; Holden 1971; Ferro and O’Neill 1994; Lowry and Robertson 1996). A cod-end composed of thin and relatively flexible twine will allow more fish to escape compared to one made of stiffer and thicker material (Boerema 1956; Isaksen et al. 1990). Thicker twines may thus decrease cod-end selectivity by increasing the resistance of the mesh openings (Lowry and Robertson 1996). Twines with lower specific gravities (e.g., synthetic fibres) tend to be more buoyant than natural fibres, which will ultimately influence the form and behaviour of the net when towed (Radhalekshmy and Gopalan Nayar 1973; Currie et al. 2019).

Around the British coast, hemp (Cannabis sativa) was largely used in the construction of trawl nets until the 1860s (Edwards 1909). However, hemp was substituted for Manila hemp (Musa textilis) in subsequent decades due to claims of increased strength (Collins 1889; Wood 1911). This pre-dates the first use of synthetic fibres in trawl nets by a substantial margin, as these only became widespread in the late 1950s (Davis 1958). Natural twines soon became a redundant component in the construction of trawl nets as synthetic twines possessed greater durability and strength (Davis 1958; Radhalekshmy and Gopalan Nayar 1973). These advantages further removed the need to constantly treat nets with dressings and preservatives (Davis 1958). Historically, hemp or Manila nets were soaked in coal tar to increase stiffness and longevity of the fibres as well as to reduce water absorption (Atkins and Warren 1953). Synthetic twines do not undergo this process as they have a higher resistance to rot and mildew (Online Resource 1).

Historically, trawl nets were braided by hand as machine-made nets could not be adapted to the shape and taper required for fishing (Alward 1932; March 1953). In the early days of trawling, nets were typically made at sea by the smacksmen operating the trawl vessel but were later hand-woven by women ashore (March 1953). Following increased demand for trawl nets in the late nineteenth century, twines were knotted around pieces of hardwood or rounded ‘spools’, which were shaped into sets to standardise the sizes required (Alward 1932; Davis 1958). Today, machine-made synthetic trawl nets are set in looms based on the desired twine material, thickness and mesh size, and mass produced in sheets for commercial use (personal communication Darren Edwards, Brixham Trawl Makers Limited).

Reconstruction

Materials used in the original gear were not specified in NSFC reports; however, it is likely that the net was composed of Manila hemp based on its widespread use around the British coast in the late nineteenth century. As natural twines have now almost completely been replaced by synthetic netting, Manila hemp was not readily available in sufficient quantities for the reconstructed net. We therefore chose white nylon as the next best alternative based on a twine that was logistically available to us and recommendations by a British net manufacturer (personal communication Darren Edwards, Brixham Trawl Makers Limited). Like Manila hemp, nylon fibres produce a relatively stiff rope but possess a slightly lower specific gravity (1.14), whereas polyester has the same density (1.38) and has softer fibres relative to Manila (Online Resource 1). In contrast, polyethylene nets float in water and have a much lower specific gravity (0.95) than Manila.

For the replica trawl, we used machine-made netting due to the high costs and difficulties associated with obtaining hand braided net panels. For the construction of modern trawls nets, the standard procedure is to order large quantities of netting (sheets) in a given mesh size (personal communication Darren Edwards, Brixham Trawl Makers Limited). Fortunately, we found a netting supplier (Swan Net Gundry Ltd) who could provide exact quantities of netting required for each net panel (based on our design plan) and reset their looms to our desired mesh sizes (see Mesh sizes for details). Unlike natural twines, our replica net was not treated with any coatings or preservatives.

Mesh sizes

Review

There is no universal protocol for describing a trawl net mesh size, but they have been described in a number of different ways since the nineteenth century. Mesh size nomenclature was originally based on “finger-widths” of trawlermen at sea, starting with four to five finger-widths in the square of the net and decreasing to two in the cod-end (Alward 1932). Meshes were later hand braided ashore and wooden ‘spools’ were used to standardise their size (Davis 1958). Today, the mesh size of a modern trawl net is often described by the longitudinal distance of two opposing knots when fully extended. This tends to refer to the distance within the mesh excluding the knots (Currie et al. 2019).

While most authors are explicit in the way they refer to a mesh size, misinterpretation can arise due to the range of ways they are described. For instance, a mesh of 2 inches was often described as 2-inch ‘bar’ or ‘square’, both of which could refer to a measure along one side of a mesh, or denote the distance between two opposing knots of one mesh when fully stretched (i.e., 4 inches extended longitudinally), or it could be the sum of all four sides (Holt 1895; Scofield 1948; Davis 1958). In large commercial trawl nets (beam and otter), mesh sizes in the upper part of the net (“square”) typically ranged from 15.2–20.3 cm fully stretched (3–4-inch ‘bar’), grading down in size through to the baitings and cod-end (Holdsworth 1874). On the other hand, cod-end meshes were typically much smaller at ~ 7.6 cm fully stretched (1.5-inch ‘bar’; Online Resource 1). A cod-end was frequently braided with double twine and the rest of net was typically constructed using single twine (Holdsworth 1874).

Reconstruction

Mesh sizes were not estimated from historical photographs due to the lack of resolution. Combining evidence reported in the historical literature, the minimum size of fish species recorded during NSFC trawl surveys (~ 10 cm), and mesh availability, we chose a 3 mm double twine with a 75 mm mesh size (inside mesh) as being appropriate for the cod-end. For the square, 104 mm mesh (inside mesh) with a 4 mm single twine was used, whereas the remainder of the net was made of 90 mm meshes (inside mesh) with a 4 mm single twine.

Trawl heads

Review

The trawl heads, or head irons, were large, metal frames that functioned as runners to raise the beam off the seafloor, approximately 91–121 cm in large beam trawls ground (Collins 1889; Wood 1911; March 1953), and to keep the mouth of the net open to allow fish to enter (Fig. 2; Collins 1889). In late nineteenth century beam trawls, trawl heads were stirrup-shaped, however their design varied at different ports along the British coast (Holdsworth 1874; Collins 1889). The most common frames, used by trawlers from Grimsby, Hull and other major fishing ports on the eastern coast, were those illustrated in Fig. 2a. In this design, beams were squared off at the end and fixed into sockets above the top part of the trawl head (Holdsworth 1874; March 1953). In less common designs, either the sockets or the iron loop through which the ground-rope passes (or both) were constructed on the inside of the frame (Fig. 2b; Holdsworth 1874). The latter was often used by smaller vessels that trawled inshore at ports such as Lowestoft and Ramsgate, whereas a more semi-circle design with sockets fixed on the inside of the upper part of the trawl head, known as the ‘Barking pattern’ (Fig. 2b), was frequently adopted by trawlers in the Thames and Yarmouth (Holdsworth 1874; Collins 1889).

Fig. 2
figure 2

Adapted from Collins (1889), which is in the public domain

Drawings of late nineteenth century beam trawl heads, showing a) a stirrup-shaped design commonly used in eastern ports of England, attached to a wooden beam and net, and b) a design that was often used by inshore trawlers (upper drawing) and a ‘Barking pattern’ (lower drawing) employed by vessels in the Thames and Yarmouth (lower drawing).

Full size image

The size and weight of trawl heads employed in the late nineteenth century varied according to the size of the net and beam as well as the depth of water it was worked in (Collins 1889; March 1953). Unlike otter boards, exact dimensions and weights of late ninteenth century trawl heads were rarely reported in the historical literature. Instead, the overall weight of a pair of large trawl heads were described. The most commonly referenced (minimum) weight for a pair of trawl heads used in a beam trawl of about 14 m was ~ 101 kg (Holdsworth 1874; Collins 1889; Davis 1927; Alward 1932; Butcher 1980). However, on large North Sea sail trawlers, the weights of the trawl heads were much larger, ranging from 160 to 181 kg (Collins 1889). Constructed from wrought iron, the frame of the trawl heads was the same thickness, except the trawl ‘shoes’ at the bottom were often made twice as thick in order to compensate for the abrasive action of the seabed (Collins 1889; March 1953; Butcher 1980).

Reconstruction

Based on materials available, we reconstructed the trawl heads using black steel, which has a similar specific gravity to wrought iron (~ 7.8; Reade International Corporation 2018). The weight of the two trawl heads were approximately 100 kg; similar to the minimum weights reported in the literature (Holdsworth 1874; Collins 1889; Davis 1927; Alward 1932; Butcher 1980). Fortunately, one of the trawl heads used on board the Livingstone is clearly visible in a photograph of the crew hauling in the net (Fig. 3) and resembles the most common trawl head design employed in eastern British ports (Fig. 2a). After measuring and scaling the dimensions from the image, we estimated the trawl heads to have a height of 68.6 cm, which is about a third smaller than the heights reported in the literature. The depth and length of the trawl shoes were assumed to be 1.3 and 115.8 cm, respectively.

Fig. 3
figure 3

Photograph of the original 6.7 m beam trawl being hauled by members of the NSFC on the starboard side of the paddle tug Livingstone (Meek 1899b, which is in the public domain)

Full size image

For the reconstructed trawl heads, we used a similar square socket to the one visible in Fig. 3. The socket was bolted to the main frame using 12 mm bolts at d to provide a flat union between the socket and the main frame a and bolted in the centre at c. The trawl shoes were stitch-welded to the base of the frames and secured at the heel of trawl head (b), resulting in a total depth of 2.6 cm (twice the thickness relative to the main frame). For attachment of the bridles, a 16 mm eyebolt was shackled into the front of each trawl head. The positioning of the eyebolt is an important setting for the effectiveness and efficiency of the trawl heads as the gear is it towed over seabed (personal communication Nigel Gray, Blyth Tall Ship). A detailed plan of the trawl heads is provided in Online Resource 2.

Ground rope

Review

Running along the lower edge of the wings and bosom in the centre, a ground-rope, consisting of an old piece of ‘towing hawser’ rounded with a smaller Manila or hemp rope (Fig. 4), served to disturb the sediment and entice bottom fish into the net (Holdsworth 1874; Collins 1889; Wood 1911). In large beam trawls, Collins (1889) reported that these ropes were approximately 60 mm diameter (7.5 inches in circumference) and 16–20 mm in diameter (2–2.5 inches in circumference), respectively. Other authors noted that ground-ropes had a central core of wire that was covered with old netting and rounded in the same way (Kyle 1903; Wood 1911; Davis 1927). The smaller rope was “rounded” from end to end to increase the overall weight of the ground-rope and to prevent it from chafing during contact with the seabed (Fig. 4a; Holdsworth 1874; Collins 1889). Ground-ropes were attached to the meshes net via a ‘bolsh’ or ‘balch’ line (Fig. 4b), while a short piece chain connected either end of the ground-rope to the back of each trawl head (Collins 1889). Bolsh lines were approximately 1 cm (1.25 inch) in diameter and set at 30 cm increments along the wings and at smaller distances along the bosom (Collins 1889; Kyle 1903; Davis 1927).

Fig. 4
figure 4

Drawings of how late nineteenth century ground-ropes a) were rounded and tickler chains attached (adapted from Butcher 1980), and b) were set to the net meshes via the bolsh line (adapted from Davis 1927)

Full size image

As in modern nets, historical ground-ropes varied in length and diameter depending on location, target species, net size, and habitat over which it was worked (Davis 1927). In 14–15 m beam trawls, ground-ropes were typically 28–29 m in length (Collins 1889; March 1953). Moreover, it was common practice for trawlers in the late nineteenth century, particularly beam trawlers targeting flatfish species such as sole (Solea solea), to further weight their ground-ropes with heavy lengths of chain to increase contact with the seabed (Collins 1889; Kyle 1903). The arrangement of these ranged from a piece of chain (4.5–6 m in length) or leaded weights attached the middle of the ground-rope (Holdsworth 1874; Collins 1889; March 1953) to a series of metal rings spaced in equal increments around the ground-rope, each linked by an iron chain (Kyle 1903).

Reconstruction

The ground-rope used by the NSFC during scientific trawl surveys is visible in photographs of the crew hauling in the beam trawl (Fig. 3 and 5), which appears to be consistent with the aforementioned description. After scaling the photograph, we estimated the ground-rope to be ~ 70 mm in diameter. This estimate was smaller than the ground-rope diameters reported for larger beam trawls. For our replica ground-rope, we used a 36 mm diameter braided staple rope rounded with a 16 mm diameter rope (Online Resource 2), which resulted in total diameter of ~ 68 mm. The chain used to connect the ground-rope to the trawl-heads and part of the bolsh line is also visible in Fig. 3, albeit the chain appears to continue along the underside of the ground-rope. This implies that a chain was either used around the entire length or placed in sections of the ground-rope, instead of just in the bosom area as reported in the historical literature.

Fig. 5
figure 5

Photographs of NSFC crew hauling the beam on the starboard side of the Livingstone by hand via a central bridle and the fore-and-after bridles (Meek 1899b, which is in the public domain)

Full size image

For the reconstructed ground-rope, we attached short pieces of chain in sections on the underside of the ground-rope. A short piece of iron chain was used to attach either end of the ground-rope to the trawl-heads through the iron loop (Online Resource 2). To connect the ground-rope to the meshes, we used a bolsh line that was set in larger increments along the wings and smaller in the bosom. Following initial gear trials in June 2018, we extended the ground-rope by 1 m either side of the wings to compensate for slack in the bolsh line, resulting in a total length of 14.5 m.

Beam

Review

The size of a beam was largely determined by the length and power of the fishing vessel that towed it (Holdsworth 1874; Collins 1889; Alward 1932), but also by the vessel’s capacity to adequately stow the beam (Holdsworth 1874). In the late nineteenth century, inshore trawlers worked small beams from 3 m upwards (Collins 1889), whereas large vessels, generally for deep-sea trawling, could tow beams up to 14–15 m in length (Butcher 1980). The diameter of a beam ranged from 20–22 cm (Butcher 1980). Historically, beams were made of wood, typically ash, beech, oak or elm; the latter being the most common (Holdsworth 1874; Collins 1889; Aflalo 1904; Edwards 1909; Wood 1911). A single piece of wood was usually employed in beams up to about 10 m, however beyond this size, two or sometimes three pieces of wood were scarfed together and reinforced with iron bands to form a long splice (Collins 1889). This method was used to increase the strength and durability of larger beams (Davis 1927).

Reconstruction

For the replica gear, we initially tried to imitate the historical beam as closely as possible to match the length (6.7 m) using a single piece of timber. The beam was shaped into a 14 cm2 square socket at end of each trawl head, as estimated from Fig. 3. Due to the lack of beam specification, we compared the densities of currently available timber with the densities of timber known to be used historically, to suitably select a material for our reconstructed beam. We chose Douglas fir, largely because it has a relatively similar density (530 kg/m3) to English elm (550–600 kg/m3). Upon testing the gear during initial gear trials in June 2018, buoyancy issues arose as the wooden beam prevented the net from settling on the seafloor. Collins (1889) suggested that spare beams were sometimes soaked in water for several hours before use in trawl ports. Other than this, there is no evidence to suggest that beams were sunk prior to trawling (or for how long). Unlike the net, the materials used to construct the beam is unlikely to influence fishing performance (personal communication Neil Armstrong, Blyth Marine Station). To counter the buoyance issues, we subsequently used a 6.7 m steel beam with a diameter of 15.2 cm in place of the wooden beam.

Methods of operation and towing speeds

Review

In the late nineteenth century, smooth fishing grounds and a favourable tide were essential factors for sail smacks and paddle tugs. The shooting and hauling of a 14–15 m beam trawl was a laborious task and manoeuvring it into the correction position required careful skill (Wood 1911). Beam trawls were towed over the seafloor by a single trawl warp, consisting of a hemp or Manila rope, attached at one end to two smaller ropes, the bridles, which were fastened to the front of each trawl head via swivel eyebolts (Holdsworth 1874; Collins 1889). Firstly, the cod-end was thrown over the port side of the vessel, followed by the rest of the net, until the net was cleared of the vessel and trailing from the beam (Holdsworth 1874). The beam was then lowered over the side, carefully maintained in its upright position, and net streamed away from the vessel at sea surface. The fore-bridle was slackened to allow the fore trawl-head to ‘square away’ at a 90° angle to the stern of the vessel, while the other trawl head remained in place by the dandy bridle, which was secured to the trawl warp at one end and the aft end of the beam (Holdsworth 1874). The bridles and warp were then ‘paid out’ until the trawl settled on the bottom at a distance astern and windward of the vessel (March 1953). Once the required length of warp was reached, the warp was secured through a stopper and towing post, which helped control the rate at which the warp was paid out (Davis 1958).

On the windward side of the vessel, trawl nets were hauled in using the motive power of steam-powered capstans and winches, which lightened the load of the crew hauling in the catch (Collins 1889; Butcher 1980). At the junction between the warp and bridles, the dandle bridle was cast off the trawl warp and later attached to a dandy winch, which was used to heave the after end of the beam towards the stern (Collins 1889; Davis 1958; Butcher 1980). The after bridle was heaved in first and secured to the taff-rail, followed by the fore-bridle, which was bought in via a forward winch (Collins 1889). As both trawl-heads swung in line with the port side, they were lowered and secured on deck, while the remainder of the net and ground-rope were hauled in by hand (Collins 1889; Davis 1958). Next, the ‘cod-line’, extending from the forward end of the beam to the cod-end was used to haul in and suspend a heavily weighted cod-end over the deck via the capstan. This was to relieve the strain of the rest of the net before the ‘poke line’ was untied to release the contents of the net (Collins 1889).

From the various scenes captured of the crew hauling in the original beam trawl onboard the paddle tug in 1889 (Fig. 5), it appears the gear was hauled in manually from the starboard side of the vessel. There was no evidence of any mechanical winches or a capstan on the Livingstone to aid with retrieving the net (see Online Resource 1). Instead, a third bridle, attached in the centre of the beam, was used in addition to the fore-and-after bridles to haul in the net (Fig. 5). A larger, more powerful paddle tug, Stanley, was later (from 1901) employed by the NSFC for scientific trawl surveys, which was 33 m in length, 5.7 m wide, and 3 m draft (Meek 1901). Meek (1901) stated that a ‘powerful steam winch and davits’ were used, which made hauling in the trawl lighter work for the crew.

Between the nineteenth century and today, the process of shooting and hauling a trawl net became more mechanised and efficient, which vastly enhanced the fishing performance of trawlers over time (Engelhard 2008). Steam trawlers were less powerful and slower compared to modern vessels (Currie et al. 2019), varying among different classes of trawlers and gears. Trawl speeds were not systematically documented in historical reports by the NSFC, however Meek (1901) reported that ‘Unless otherwise stated, the steamer was trawling at the time the net was being used, and the speed would therefore be about 2 knots.’ This speed is supported by similar towing speeds reported in the literature from the same era. For example, Garstang (1905) obtained 2.1 knots on average during trawl investigations conducted on the SS Huxley in the early twentieth century. Kyle (1903) reported that steam-driven beam trawls were towed at 2 knots, while M’Intosh (1895) reported that the average towing speed was ~ 2.5 knots. Moreover, a sail smack could reach speeds of 8–9 knots, but this was offset by 6–7 knots due to the drag of the net (Holdsworth 1874).

Reconstruction

For our repeat surveys, the reconstructed beam trawl was deployed from the stern of a 18.9 m catamaran, The Princess Royal (see Online Resource 1). The vessel has an overall width of 7.3 m and draft of 1.6 m. In modern trawl vessels, it is common practice for trawl nets to be operated from the stern to enable faster shooting and hauling of the net, further reducing time between trawls as the cod-end is emptied and reassembled while the vessel is repositioned for the next trawl (Currie et al. 2019). We shot and hauled in the replica beam trawl using a single cable warp attached to two cable bridles at one end, which were in turn connected to the front of each trawl head via an eyebolt. The length of warp used for trawling was controlled by a central trawl winch via a hydraulic A-frame at the stern of the vessel.

During gear trials, we maintained a towing speed of 2 knots (1.03 m/s−1). This was based on trawl speeds reported in the literature and speeds calculated from tow durations and approximate distances trawled during historical NSFC trawl surveys. The mean towing speed calculated was estimated to be 1.8 knots (0.93 m/s−1) in historical NSFC trawl surveys.

Gear trials

In all, a total of 20 hauls were made using the replica beam trawl and 8 with the otter trawl in August 2018 (Online Resource 3). In March 2019, we conducted a further 6 trials using the replica beam and 4 with the otter trawl. Of the 6 trawls made by the replica beam trawl in March, three hauls yielded no fish whatsoever. In August, similar median catch rates were observed in dab, plaice and flounder (Platichthys flesus) between the two gears, albeit medians were slightly higher for the latter two species caught by the otter trawl (Online Resource 3). In comparison, the median catch rate of European lobster (Homarus gammarus) was substantially higher for the otter trawl than the beam trawl. By contrast, median catch rates of all fish species landed by the beam trawl in March trials were either 0 or marginal, whereas substantially more dab and plaice were landed by the otter trawl in this period, although still much lower in August. Wilcoxon Mann–Whitney U-tests detected significant differences in flounder and lobster catches between gears in August trials (P < 0.001 and P < 0.05, respectively), but not for flounder (P = 0.065). Conversely, Mann–Whitney U-tests revealed significant differences in dab and plaice catches (P < 0.05) but not for flounder (P = 0.236) between the two trawl gears in March.

In August trials, the mean lengths were significantly smaller in the replica gear for plaice (Student’s t-test, P < 0.001) and dab (Welch’s t-test, P < 0.001) compared to the otter trawl. Similarly, the mean lengths of these two species were smaller in March gear trials (Online Resource 3) but were not tested statistically due to low sample sizes in the replica gear for this period. A Kolmogorov–Smirnov test also revealed a significant difference in the length-frequency distribution of plaice between gears (D = 0.39, P < 0.001) but not for dab (D = 0.39, P = 0.131).

Discussion

This study provides an important repository of information for understanding the design, technology and operation of beam trawls employed in the late nineteenth century. Information was compiled from a wide range of historical sources, including reports, literature, photographs, and newspaper articles, as well as input and advice from numerous stakeholders within the trawl industry, demonstrating the feasibility of reconstructing a beam trawl that was first used in scientific trawling investigations over a century ago. On this basis, we provide a detailed design plan, incorporating features and dimensions that were carefully scaled to fit a 6.7 m beam, which aimed to functionally replicate a historical beam trawl using contemporary materials (summarised in Table 1). Following construction, we tested the performance and catchability of the replica beam trawl by conducting a series of comparative trawls (with a modern otter trawl) in August 2018 and March 2019. The otter and replica beam trawl exhibited broadly similar efficiencies in catching flatfish during the August trials. In March, however, the otter trawl was more efficient at catching flatfish (albeit still much lower than in August) in contrast to the replica gear. During both trial periods, the mean lengths of dab and plaice landed by the replica gear were smaller compared to the otter trawl.

A complete picture of the dimensions and how the original gear was constructed was not possible, thereby requiring several assumptions to be made and uncertainties to be carefully addressed. For example, there was a lack of detailed information on late nineteenth century beam trawl plans that were similar in size to the one used by the NSFC. Instead, the reconstructed net based was on plans and descriptions of large commercial trawls with ~ 14–15 m beams that were towed by sail and steam-powered vessels in the mid to late nineteenth century (Holdsworth 1874; Collins 1889), but scaled-down to 6.7 m. The paucity of information is surprising considering that primitive beams were ~ 3–4 m in length and were in widespread use around the British coast from the mid-eighteenth century onwards (Robinson 1996; Jones 2018). Moreover, it is likely that variations in gear design were minimal as significant modifications were not adopted until the invention of the otter trawl in 1894 (Cunningham 1896; Kerby et al. 2012).

Where historical materials were not available, we faced some challenges with sourcing appropriate contemporary materials for the reconstructed net. Compared to hand-braided nets made of natural fibre twine, modern trawl nets are machine-made and consist of synthetic fibres, which have superior quality (e.g., breaking strength and durability; Davis 1958; Kerby et al. 2012) and cheaper production costs. Today, it is standard practice to supply netting in bulk in a specific mesh size (personal communication Darren Edwards, Brixham Trawl Makers Limited). Compromises were therefore made during our attempt to reconstruct replica trawl net. Properties such as flexibility, specific gravity and elongation at break can also influence the potential escape of fish as a function of their size and shape (Isaksen et al. 1990; Lowry and Robertson 1996). Compared to the stiff, natural fibres of the original net, it is possible that fewer fish may have been retained by the synthetic replica net as nylon exhibits a greater elongation at break (10–12% versus 15–28%, respectively). Some studies showed that nylon-made trawl nets landed fewer fish (10–15% less) compared with similar mesh nets made of Manila twine (Clark 1956, 1963). Nevertheless, white nylon produces a relatively stiff rope and was selected as the next best alternative that was logistically available. In November 1998, an attempt was made by fishery scientists and members of the Excelsior Trust in Lowestoft to trial a replica 1880’s beam trawl to investigate the fishing power of an old sail trawler, the Excelsior. They also used white nylon to reconstruct the net, based on materials available that most matched the appearance and density of natural fibre twines (personal communication, John Wylson, The Excelsior Trust). In South Africa, Currie et al. (2019) used a combination of polyester and polyethylene for the reconstruction of a late nineteenth century ‘Granton’ otter trawl net made of Manila hemp. They used polyester because it has the same specific gravity in water as Manilla (1.38) and polyethylene to imitate a similar stiffness, to produce a rope that, together, would exhibit a similar fishing performance to the original Granton net. In our study, it was not feasible for the net manufacturer to use this specific combination because synthetic nets are typically manufactured in standardised sheets on a mass scale for commercial use only (personal communication Darren Edwards, Brixham Trawl Makers Limited).

The reconstructed beam trawl underwent various modifications following initial testing in June 2018 (see Table 1) and prior to comparing the catch efficiency of both modern and historical trawl gears. Firstly, the wooden beam was replaced with a steel beam due to buoyancy issues, which prevented the trawl from settling on the seafloor. Secondly, we reduced the number of meshes in the front-width of square of the trawl net from 180 to 160 and lengthened the ground-rope by 1 m either side of the wings. Prior to this second adjustment, the gear did not trawl for fish effectively and we subsequently ‘streamed’ the net alongside of the vessel whilst moving to visually assess the behaviour of the net at the sea’s surface. It became apparent that there was excess netting in the upper portion of the net that caused the net to “balloon” as it was being towed at the surface. Consequently, we addressed this issue by slightly reducing the size of the square and extending the ground-rope to compensate for slack in the bolsh-line (see Table 1).

Although the replica beam trawl had a similar catch efficiency to the otter trawl (for flatfish) in August 2018 after being adjusted, catches from the replica trawl gear in March 2019 were negligible. In fact, three out of the six hauls had zero catches. Consequently, the replica beam trawl was deemed to be not functioning efficiently in this period. It was therefore not used for its intended purpose of resurveying the same bays as the NSFC did to investigate long-term change in the inshore fish community. We were unable to investigate or resolve these issues further due to practical and logistical constraints of the study. There are several factors that could explain a) the lack of functionality of the replica beam trawl and b) differences in the catch rates between the beam and otter trawl during gear trials. These are unpacked in more detail below.

For more than a century, trawl gears have undergone significant improvements in their design, materials and methods of operation (Engelhard 2008; Currie et al. 2019). For example, their hydrodynamic design has become more streamlined to optimise fishing performance and reduce drag, specifically for the intention of improving selectivity and minimising their physical impact (Burgaard et al. 2023). Because the specifications of the replica beam trawl were based on late nineteenth century net plans, the gear was not designed to be streamlined or to improve catch efficiency. This could have conceivably impeded the gear’s performance during the comparative trials, but it also underpins the “ballooning” effect observed in the square during initial trials in June 2018. Despite reducing the number of meshes to address this, it further highlights the lack of optimisation in the hydrodynamic design of late nineteenth century beam trawls.

It is possible that our results could partly reflect differences in the catchability of the two gears. Trawl geometry, variable for the otter trawl and fixed in beams, as well as the degree of contact between the ground-rope and seabed, can influence catchability in terms of species composition and size (Jennings et al. 2001). However, this variability was minimised by the fact that all the gear trials were conducted in depths of < 25 m. To further account for possible differences in catchability between the two gears, we applied appropriate swept area calculations and excluded smaller fish < 15 cm (total length) from our analyses. Although these do not fully resolve differences in catchability, both trawl gears exhibited relatively similar catch rates for flatfish in August 2018 and neither landed any roundfish. This further supported the applicability of using the otter trawl in a later study, where we contrasted contemporary fish catches with historical data collected from the original NSFC beam trawl (see Hunt et al. 2024).

Compared to the August 2018 trials, a higher proportion of flatfish were landed by the otter trawl in March 2019. In part, these differences could be attributed to the seasonal variability in their distribution patterns due to changes in temperature, food availability, and the provision of suitable spawning and nursery grounds. Since the 1980s, plaice and dab have shifted their distribution to deeper and cooler waters offshore, which has been attributed to the avoidance of increasing temperatures in inshore waters (van Keeken et al. 2007; van Hal et al. 2016). This could thus explain why fewer individuals were landed by the otter trawl in August 2018 compared to March 2019. However, it does not explain the zero or negligible catches exhibited by the replica beam trawl; we thus remain convinced that the gear was not operating sufficiently during this period.

While it was not possible to resolve the functionality of the replica beam trawl, this work is of value to researchers interested in historical trawl gear technology. To our knowledge, this is one of only two studies (Currie et al. 2019) that has set about to replicate like-for-like trawl gear in an attempt to investigate long-term change in a fish community. Although in April 1998, an initial attempt was made to investigate the fishing power of an old sail trawler, the SS Excelsior, using an 1880’s replica beam trawl, which was historically capable of landing vast quantities of fish in the southern North Sea. However, the authors of this study failed to land any fish despite their considerable efforts to imitate historical fishing practices (Millner et al. 1997; Engelhard 2008). In their second attempt, the gear was trialled twice for ca. 45 min off the coast of Lowestoft in November 1998, and was successful in catching a small number of fish species, primarily sole (S. solea) (personal communication John Wylson, The Excelsior Trust). Historically, beam trawls were manually hauled in by experienced and skilled crew without the aid of a mechanical winch to haul in the catch (Collins 1889). Although the same experienced skippers consistently handled the replica beam trawl, we used a mechanical winch at the stern of the vessel for its deployment and retrieval. It is a very challenging task to accurately imitate the same methods as the scientists and crew of NSFC did in the late nineteenth century, especially when the technology, skillsets and experience levels are very different than those employed today. Engelhard (2008) suggested that the unsuccessful trial of the 1880s replica beam trawl used by fishery scientists aboard the Excelsior may have been attributed to the lack of fishing power and skills required to handle the historical gear. Could the low catch efficiency of the replica beam trawl therefore be in part due to our lack of experience in working a late nineteenth century beam trawl? Or are fish just far less abundant compared to a 120 years ago?

We demonstrate that the combined social and technical aspects of a fishing practice in terms of vessel, trawling gear and the crew’s skillset are incredibly challenging to re-enact. These three components have each evolved over generations through practice and incremental innovations by fishing communities, and general technological developments, which have reshaped the efficiency of practices for centuries. As Edgerton’s “Shock of the Old” describes (2008), innovation can occur through the adaptation, reuse and rediscovery of historical technologies alongside novel inventions. Through history, fishers have progressively modified and repurposed existing trawl gear to adapt to different types of grounds, shifts in target species, environmental conditions, and evolving management restrictions (Kennelly and Broadhurst 2002; Engelhard 2008). Our work recognises this process of "reinvention" by acknowledging the significance of old and new trawl technology. Beam trawls originated in European waters and marked the first major development of the commercial trawl industry (Alward 1932; Visel 1980). This work therefore provides an invaluable repository for investigators of trawl technology, fishing power and survey design. It also provides an internationally important foundation for other fishery-independent surveys that used similar beam trawls around the British coast (Fulton 1895; Garstang 1905) and across the wider North Sea (Garstang 1905; Todd 1911), but also to other trawl technology employed elsewhere in the world (e.g. Norway: Hjort 1914; Australia: Department of Trade and Customs 1909; USA: Alexander et al. 1915; South Africa: Currie et al. 2019, 2020).