Opportunities with Multi-Layer Weave Structures in Woven E-Textile Design

Woven structures are categorised by the number of ‘yarn systems’ (sets of warp and weft yarns that move independently within the structure to construct the textile) and their arrangement. These two factors determine the woven structure's complexity and layer count. Simple weaves, using one warp and one weft system [21, pp. 74–90], commonly feature in clothing fabrics.

Simple weaves, the woven structure cited most frequently in e-textile literature [90], provide a stable one-layer base for integrating electrically functional materials. They can serve for sensing touch [22, 77, 115], displaying changing patterns [7, 40, 106] and carrying signals [58, 73, 84].

The simplest of all structures is the plain weave in which every thread passes alternately below and above the cross-wise direction's consecutive threads. However, several other yarn arrangements are possible, each with its own effects on the e-textile's electrical behaviour [15]. Basket [77] and twill [98] weaves, characterised by longer yarn ‘floats’ than a plain weave has, reduce compression and tension in weft and warp yarns, thereby extending the sensitivity range of sensing elements [77] and improving wearable sensors’ adaptation to body movements [98]. With even longer floats and less interlacing, a satin weave lets the fabric incorporate light-emitting-diode (LED) fibres [93]. Waffle weave is a simple weave where the yarns are at greater distances from each other. This makes it looser, giving the fabric more volume. Arranging conductive/resistive yarns in a waffle structure enables sensing pressure [28, 78] as compression changes the fabric's electrical response.

More complex than simple weaves, compound weaves make use of multiple yarn sets. Weaves produced with an additional weft are called double-faced weaves, because they permit creating two-sided fabrics [3], gradients and mixtures and two-colour figurative patterns [47, 117]. If the yarns in the two weft systems are of different materials, distinct functions can be realised. For instance, if camouflage is required [79] or skin contact must be avoided [75], one side might embody embedded functionality while the other provides insulation. More colourful and intricate patterns can be created through Jacquard weaving, which typically brings two or more weft systems into play. It supports labelled LED displays [76], various visual cues needed in touch-based interfaces [89] and other electrically functional figurative patterns.

The warp direction too may have additional yarn systems: parallel but independently controlled longitudinal yarn sets. In double weaves, two layers are created simultaneously by means of two warp and two weft sets [21]. This is the complex woven structure seen most often in e-textile prototyping [90]. When the two layers have independent weft systems, the layers remain fully separated, whereas using a single weft system gives the outcome a tubular structure, which can benefit the fabrication of 3D woven e-textile garments [32] and deformable woven patches for on-skin haptic feedback [52]. Also, the two layers can cross, forming tubes and pockets. Two-layer weaves’ hollow structures have been used to envelope electrical components [14, 28, 50] and other materials [16, 78, 89, 98] inside a textile. This serves particularly well for pressure sensors. One technique interweaves conductive yarn with each layer of a double-weave pocket while piezoresistive velostat [16, 89, 98] or conductive yarn floats [78] act as a medium between the two. Double weaves also let designers create electric-switch structures. Conductive yarns used in the separate layers remain disconnected in their normal state but form a contact upon pressing [16, 91].

Alternatively, the two layers can be interconnected (bound together by a set of warp or weft yarns crossing between them [e.g., 1, 48]) to create a structure called double cloth. More robust than double weave, this structure is especially commonplace in upholstery. E-textile circuitry between the two layers can embed signalling [2, 74], component-specific yarns [37], or both signal traces and integrated components [62], with the sturdy structure protecting against exposure of the between-layer elements and insulating the conductive materials. Double cloth even enables crafting touch-sensitive interfaces, through an embedded capacitive grid structure composed of conductive weft and warp yarns [83].

The layer count need not be limited to 2. This freedom allows for multi-layer fabrics with variously linked or separated layers and for inclusion of spacer materials [44] or multiple levels of electrical conductivity [81]. The density and thickness too can vary. Therefore, one can construct pressure sensors by sandwiching a low-density insulating layer between layers of conductive yarns [78]. Upon compression, the conductive layers come in contact and a press gets detected, with the option of spacer layers enhancing the system by increasing the sensor's volume and springiness [4, 25]. Additionally, layer-specific elasticity levels afford stretchable signal paths [34]. As elastic outer layers bend, the non-elastic layers below them may compress and wrinkle while those layers’ embedded conductive strands remain unharmed; likewise, said strands can elongate when the textile is stretched.

Many-layer fabrics also let touch-sensitive areas span several layers. Using non-conductive outer layers to protect integrated components and electrical signalling woven into the middle layer represents an inversion of such structures [63]. Three-layer structures support various other sorts of signal flows also. Embedded antennas are one example [30]. Also, electrical signalling integrated into the inner layers can be brought to the textile's surface at selected locations for post-weaving connection of external components [99].

Finally, cutting certain layers in a certain manner after weaving transforms a planar woven textile into a 3D shape, thus affording entirely new forms of interaction [11]. Multi-layer textiles’ most extreme form relies on 3D woven fabrics with dozens of layers [44]. While woven shoes [39] and various other fully woven 3D objects could be crafted with 3D woven structures, their applicability in e-textile development has yet to be explored.

Compound weaves can be created or augmented by means of supplementary yarn systems. A supplementary warp or weft is an additional system that does not contribute to the textile's fundamental structure. Traditionally these function for decoration, reinforcement, or creation of float and pile structures at the surface [21]. Thanks to supplementary yarns of a material, density and/or colour different from the base fabric's, woven e-textiles can offer additional functional, visual and tactile properties [87, 89, 112], as in surfaces that sense gestures by means of capacitive grids [87]. Meanwhile, the complexity required of the base textile is not affected.

Techniques in this family are ideal for embedding functional elements in predefined locations. In the fil coupé technique, the supplementary weft intertwines with the base structure location-specifically while yarn floats are formed elsewhere [71]. At the former, the technique can create touch- and pressure-sensitive regions of various shapes [78, 87, 89], while the yarn floats can be either cut off after weaving, to disconnect the conductive areas from one another [78, 89], or used later to attach a capacitive grid to electronics [87]. Another approach is to weave in additional materials via the inlay technique, in which a continuous weft yarn is made part of the weave at a specific location in the textile along with the ground wefts [21, p. 141]. Researchers using this technique, which is more tightly bound to the base structure, have created a woven potentiometer [111] and thermochromic displays [17, 28] embedded in a broader stretch of woven textile.

We constructed the samples manually, hand-weaving them at our university's weaving studio on two looms, with different mechanisms: a TC2 digital Jacquard loom and a Toika 16-shaft dobby loom. The Jacquard loom offered complete control of each individual yarn, thus enabling construction of more complex structures. It allowed us to fabricate samples with location-specific functional elements and yarn arrangements that require control over each warp yarn. However, some parameters were beyond our control: the warp was set to use 30 yarns/cm, and its material was always white mercerised cotton.

With the Toika dobby loom, in contrast, we could use a bespoke warp composed of cotton and conductive yarns. Because dobby looms lift the warp yarns mechanically in bundles only, we could weave samples in which the overall structure and material combinations were uniform across the textile and did not require detail-level modifications.

We designed the weave structures and the material selection exploration-specifically. In explorations 1–4, we used plain-weave-based multi-layer structures and limited the material to cotton (for Exploration 1) or to conductive and insulating yarn (in explorations 2–4). For explorations 5–7, we expanded the weave and material choices to cater for more versatile electrical behaviour and to emphasise tactile differences and visually distinct outcomes.

Although the looms were manually operated, they were computer-controlled, and the weave structures were specified in digital files. Therefore, our sample-construction cycle typically was made up of sequences of digital-construction-file preparation, material selection, observation during weaving and reflective evaluation of the outcome. After several iterations, we arrived at the desired outcome. The article's supplementary materials provide thorough description of the structures and materials.

Table 1 presents details of our two-pronged evaluation, covering the methods for the sensory tests and the basic electrical assessment both. Since haptic information is crucial for textile evaluation and design decision-making [92], we took care to choose subjective sensory-assessment methods that are commonplace for assessing the ‘fabric hand’¹ in the woven-textile domain [8, 97, 100] to judge overall impressions of properties such as sample thickness, drapability and structural stability. In addition, the first author made electrical measurements with a digital multimeter, following a procedure similar to prior work's [104, pp. 119–123], both during weaving and after removal from the looms. The evaluation focused on quality control during sample creation and was aimed primarily at error-detection. In the event of errors, iteration of the structures ensued and weaving was performed again. The evaluations supported immediate reflection-in-action for iterative improvements to patterns and material selections as weaving progressed. They also aided in identifying further research questions, for later explorations.

Comprehensive evaluation of the final samples’ electrical performance was carried out as the project neared its conclusion. For this, the second author used high-precision instruments to conduct controlled experiments on artefacts from explorations 2 to 7. In general, these focused on examining how the woven structures, the layer arrangements and the materials used in the structures influenced the electrical behaviour of the samples. The tests of samples from explorations 2 and 3 measured electrically related differences between linked and non-linked multi-layer samples. Those addressing Exploration 4 evaluated how electrical behaviour was affected by yarn systems’ locations within linked multi-layer structures. Finally, we measured the electrical behaviour of the samples from explorations 5, 6 and 7 in baseline conditions and upon interaction, to examine the effect of material choice on electrical interactivity. To present the electrical responses graphically, we used a frequency-based approach wherein Lissajous-pattern signals [101] assist in identifying impedance and semiconductive properties, such as those at the surface of steel-based yarns.

Appendix A provides the technical details of these measurements.

We observed that, overall, the layers can be kept mostly separable even in nine-layer samples, without any tools more advanced than a standard manually controlled Jacquard loom. Dividing the warp yarns across multiple layers affected the warp and the weft densities differently. Since every further layer reduced the number of warp yarns per layer, the weft floats became longer. Hence, the weft yarns could intertwine increasingly densely (as the pattern of vertical densities from X1-A/A1 to X1-H/H1 in Figure 5 attests). Unsurprisingly, using longer weft floats compromised weave stability, because this made the structure looser.

We did not find a strict upper limit for the layer count in woven textiles. However, higher ones necessitate considering structural stability. This has implications for the fabric's durability. As the weft density and yarn-float lengths increase for reason of the lower warp density, the structure's balance affects the stability of each layer. Longer floats can move relative to each other when external stress gets applied. This property is strongly related to the surface's durability and abrasion-resistance. In woven e-textiles, an unstable yarn arrangement could render functional structures less reliable, leading to unpredictable electrical behaviour. Finally, by increasing the woven textile's volume and, in turn, its elasticity, a higher layer count can protect the inner layers.

Samples with an adjusted warp distribution (X2-D and X2-E) exhibited greater stability than X2-A to X2-C and those from Exploration 1. The top layer's denser warp helped make the weft floats shorter and less likely to move when exposed to rubbing. The shorter weft floats prevented the weft yarns from intertwining densely in the top layer. With the weft density for other layers following suit, that reduced the weft's yarn count and, subsequently, the textile samples’ stiffness and thickness. In addition, since the layers were less voluminous, we found the overall structure more balanced, as is evident from the cross-section images (b2–e2 in Figure 6).

The electrical measurements from this exploration supplied reference data for later explorations, particularly comparisons related to how layers’ conductivity changes upon application of external pressure to the fabric. Data from these are presented in connection with Exploration 3's results, below.

This exploration yielded promising results: multi-layer e-textiles including not just 1–2 but even three or four conductive layers are possible. Fabrics with a dense outer layer and sufficiently insulated conductive layers could be created with as many as seven layers. A denser warp for the outermost layers ensures shorter weft floats, thus increasing the surface's resistance to wear [46]. This should prove advantageous for textiles subject to external stress and abrasion (e.g., in smart work-wear and outdoor clothing). A denser warp also results in stronger intersection of the warp and weft (they cross at more points at the top). That affords greater variety in yarns’ interlacement points and, thereby, a wider array of weave options influencing e-textiles’ surface look and feel. We returned to this factor when conducting explorations 5–7.

Linking the layers together had the intended effect: when we exposed the samples to abrasion and rubbing, the internal stitching decreased their movement relative to each other. This was expected, given that woven-textile design and construction frequently relies on interconnecting to give the product a better fabric hand [1]. We found also that the stitching reduced the movement of the weft yarns at the surface, strengthening the textile's wear-resistance. Multiple linked layers thus improved on Exploration 2's results, yielding a stabler, flatter and more drapable structure than the looser, more malleable non-linked weave created.

Inspecting crosscuts of the linked structure did reveal partial merging of layers due to self-stitching, a phenomenon that could compromise electrical insulation between conductive layers, potentially even leading to short-circuits. However, the instruments assured us that the wool layers conferred proper insulation. In comparison to the non-linked layering from Exploration 2, the samples with linked layers had better protection from external mechanical influence, and placing a weight on top did not affect the signal substantially. In the absence of external pressure, signal-conduction strength depended on the use of conductive yarn in each layer: with a 1:4 ratio of conductive to plain yarn, X3-A had the weakest signal-transmission (27.2% of the maximum voltage); X3-B and X3-C, both with a 1:2 ratio, proved better (at 30.7%) and X3-D with its 1:1 ratio reached the highest values (36.4%). All levels were higher than the corresponding ones from Exploration 2. Applying external pressure caused only minor drops in these figures, of one percentage point (1 pp) (X3-A: 27.0%, X3-B: 30.6%, X3-C: 31.7% and X3-D: 37.2%). In contrast, the non-linked layer structures from all samples in the previous exploration exhibited a noticeable reaction to the mass. Figure 8 offers a graphical summary of these differences between the linked and non-linked multi-layer samples.

The measurements also revealed that all the samples could function as capacitors; that is, a signal fed to one layer ‘jumped’ to the next, mirroring how an alternating current signal behaves when passing through a plate capacitor. Capacitive performance improved as a function of the proportion of conductive yarn in the charged layers. Considering the values alongside those from the non-linked multi-layer samples showed us that Exploration 3, with its linked layers, led to better capacitive performance.

The findings point to possible stability gains from interconnection of woven textiles’ layers. This effect is especially evident in samples whose surface-layer warp density has been tweaked simultaneously. Therefore, we concluded that linked multi-layer weaves can be appropriate for interactive furniture and other durability-demanding e-textile applications. Since linking the fabric's layers entails stitching, (of which self-stitching that we used is a variant), designers must account for the effects on insulation between the inner layers, however. They must select the intersection points carefully on the basis of the conductive yarns’ arrangement. Future efforts to alleviate the problem could encompass other stitching techniques; in one option, centre stitching [1], an additional warp or weft set is reserved for stitching alone, though this presents the drawback of adding complexity to the weaving process.

The electrical results too showed promise. Importantly, signal strengths were practically unaffected by our external pressure. Our measurements also attest to linked multi-layer structures’ utility in creating new types of capacitive sensors. For example, they might hold value for measuring humidity and for detecting changes in textile materials’ dielectric properties (e.g., their response to ultraviolet radiation).

This exploration homed in on a means by which a signal from an inner layer can be brought to the textile's surface and vice versa. It combines the use of an additional conductive warp system, swapping of conductive yarn systems, and fil coupé weaving. While the technique reduced the top layer's warp density (since it requires cutting the conductive warp yarns and leaving them floating), it did not appear to degrade the fabric's structural stability, as examined via visually inspecting the crosscut and testing by rubbing. Comparison between samples X4-A and X4-B showed that the conductive warp can safely be left floating when not needed for weaving conductive areas; structural stability did not decline noticeably.

Electrical measurements of both samples when external pressure was applied (see Figure 10) revealed that the galvanic connections within the mesh of conductive weft and warp yarns were solid enough to ensure electrical connection both within each conductive region and between conductive areas on separate layers. In other respects, the measurement results were similar to those from Exploration 3: Without a mass, the voltage responses of X4-A's two overlapping layers were within 0.3 pp of each other and X4-B's responses differed by less than 0.1 pp. When the mass was applied to the input layer, X4-A showed a slight increase in both cases (below 3 pp), and applying the mass to the measurement layer produced a slight decrease in both cases (again, ¡3 id=”671” pp). Sample B, on the other hand, exhibited a slight increase (¡1 id=”672” pp) when the weight was placed on the measurement layer and a more significant increase (5 pp) upon placement on the input layer. These differences may be explained by the additional conductive layers in the first sample, which add to the electrical loading of the signal. In sum, measurements attest that the changes to the double-cloth structure neither caused the signal to deteriorate nor compromised the signals’ tendency to couple between layers.

Directly accessing multi-layer weaves’ inner structure can be difficult in interconnected textile structures; however, the exploration uncovered a mechanism to bring those signals to the surface. Our findings suggest that the fil coupé technique permits weaving multiple distinct functional structures with conductive yarns planted within the warp. Since there is no limit to the length of floats, they can isolate conductive areas within close proximity and, just as well, form functional elements across a wider surface expanse. Fil coupé does need careful planning, however. As the yarn cuts reduce the top layer's warp count, it also influences its structure and durability. Cutting too many warp yarns, especially when a wide surface area is involved, can decrease resistance to wear. At the same time, sparsely spaced conductive warp yarns form weaker connections to the surface electrode. Therefore, successful use of the technique requires finding a balance that keeps the weave structure sufficiently stable while using a dense enough conductive warp to form an electrically well-unified mesh of conductive weft yarns. A further option to explore involves using not a conductive warp to swap the locations between layers but conductive weft yarns, carrying the signal to the surface in the weft direction. That might enable more sophisticated electrically functional structures and interface layouts in which physical dimensions can be scaled (up and down) in the warp and weft direction alike. This could afford dimensioning for garments: producing functional regions proportional to the e-textile's overall size.

Furthermore, an additional option worth exploring involves placing the conductive warp on another warp beam, distinct from the ground warp. This advanced approach, facilitated by appropriate machinery commonly utilised in full-scale production, permits an increase in the yarn count of the conductive warp. Consequently, the conductive warp can be subdivided into smaller sets of separate yarn systems capable of moving independently from one another without compromising the density of the base textile. This approach could enhance control and flexibility in the weaving process while preserving the integrity of the base textile.

This exploration proved successful: the double-faced weave allowed a whole host of materials with distinct properties to merge in the top layer's two sides. Adjusting the densities of individual weft systems supported improving the combined effect. More specifically, the adjustability of individual weft systems’ and warp systems’ densities let us readily incorporate weft materials of varying thicknesses: we could design distinct layers that were fairly independent from one another. The higher warp density of the outermost layers permitted structurally more versatile weaves and afforded us greater freedom to experiment with varying textures and materials—confirming our conclusions (in Section 4.2.2) about a denser warp's implications for textile design expression on the top surface.

Having an inward-facing conductive weft rendered the functional structure more robust and the conductive yarns less susceptible to effects of rubbing. Mutual sensory comparison of the samples’ crosscuts indicated that, irrespective of the uppermost layer's tactile and visual properties, there was no significant influence on the inner functional structure.

Our electrical measurements verified this impression. The samples behaved consistently, with the surface layer's weave pattern and yarn material having little effect on the results. The signals leaving the samples were 2.5% of the input signal's level, rising to 2.8% when the external mass was placed at the measurement junction (see Figure 12, pane c). When all influences, including differences in materials and in the positioning of the external mass, are taken into account, the signal values for each respective sample lie less than 1 pp from the uninfluenced signal.

For woven e-textiles’ versatile functionality but also for rich aesthetic opportunities, the ability to design the two sides of a double-faced weave independently is highly desirable: each side could have its own electrical and/or sensory characteristics. We were successful in this regard and could free the fabric's top and bottom surfaces to express weave-pattern and weft-material variations. Our solution contributes to user comfort, aesthetics and durability in applications for interactive clothing, for example. Furthermore, we observed that our broad spectrum of surface-layer materials and weaves displayed only a minor influence on the functional structures deeper in the structure. That implies untapped potential for textile-based interaction. When constructed in the manner described above, the fabric can keep the sensing elements for its touch-based interfaces internal to the woven structure while aesthetic properties are designed autonomously.

As the previous exploration's success with a sensing-matrix structure of this general sort might suggest, we expected that a similar solution might be possible for the middle layer. Exploration 6 confirmed its buildability. The inner functional structure could be modified without the material selection or weave-pattern design for the outer layers being affected. When we adjusted the weft densities to factor in the differing weft-yarn thicknesses, the layers could be designed reasonably independently from one another.

As the nature of the material's pressure-sensitive conduction suggests, the measurements with the external mass were very different from those without it when we used piezoresistive material in the middle layer. The signal value obtained at the measurement junction was 11% of the input level (pane c in Figure 12 depicts the measurement setup) when no mass influenced the textile. When the mass was placed directly at the measurement junction, the signal value rose sharply to 91%. This change was significantly greater than the mass-free condition's and around 260 times larger than the change when cotton was used. With the mass placed at neighbouring junctions, away from the measurement junction, the value was 9%–12%, depending on the mass's exact position.

The possibility of combining layers of different materials implies that one can change the functional structure and its electrical behaviour without influencing the look and feel of the outermost layers, and vice versa. This implies that varying the material and/or the weave structure need not affect the weaving setup. Another advantage for woven e-textiles’ production is that these variations can optimise the sensing range of touch- and pressure-sensitive functional weaves for specific use cases. For example, the capacitive and piezoresistive sensors used in explorations 5 and 6 differ immensely in their measurement ranges, indicating that the middle material can be selected in line with the electrical responses required.

In the final exploration, our structurally advanced experimentation demonstrated the suitability of both weft-patterning techniques (fil coupé and inlay) for increasing the layering locally within a woven e-textile. These techniques increased the number of individual weft systems available, and adding conductive and piezoresistive weft yarns enabled us to limit electrical functionality to specific areas of a textile. Importantly, the incorporation of the further weft systems notwithstanding, the one-layer base fabric's weft density – and, consequently, its appearance and texture—remained consistent throughout the textile. Whereas the overall weft density in earlier explorations was influenced primarily by the natural settling of the yarns during weaving, Exploration 7 spotlighted the ability to control additional layers’ and weft systems’ density at base-fabric level. This opportunity is elaborated upon in Section 5.2.1.

When comparing the samples’ crosscuts and unravelling their warp and weft yarns, we found that the main differences between the two techniques lay in how they influenced the resulting structure's stability. In inlay, using an additional weft to form a single conductive inner surface for one multi-layer element culminated in a tight structure. With fil coupé, in contrast, the cutting divided the continuous conductive weft into shorter segments. This made it more prone to movement effects and less resistant to abrasion; however, we observed that the base textile's double-faced weave and its two weft systems’ division (to face both outer surfaces of the multi-layer areas) allowed for the construction of fully sealed functional weaves. Since the conductive yarns were facing inward, as the yarn arrangement presented in Figure 14 (b4 and g–h) attests, they became enveloped within the textile. The conductive yarn segments were not directly exposed to stress from outside and remained secure within the structure.

Visual inspection of the crosscuts revealed that, while cutting the fil coupé floats left the conductive weft-yarn fragments unconnected, the conductive warp aided in forming a mesh of connected conductive weft and warp yarns. We recommend, then, that any use of fil coupé for forming conductive areas should be accompanied by conductive warp yarns intertwining with the weft.

Electrical measurements showed responses that were similar to those produced by the previous exploration's sample X6-A but stronger. When measured with the multimeter and a 100 kHz signal, the samples resembled normal piezoresistive sensors, and the textile reacted strongly to external forces. For example, the resting-state value from X7-A was 48% of the input signal and adding the mass brought the level to 87% of the input signal. Two factors affected the samples’ electrical response—the surface area of the piezoresistive material and the compression caused by the woven structure itself: Mutual comparison of all piezoresistive samples from the explorations indicated that the size of the pressure-sensitive area (and hence the amount of piezoresistive material) influences the reactive behaviour's strength. Secondly, multimeter measurements (from the final samples and some earlier-iteration versions, with denser weave patterns) showed that shorter weft floats caused the piezoresistive weft material to compress somewhat, thus reducing the sensing structure's sensitivity (cf. Parzer et al. [77]). In addition, interconnecting the layers transmits compressive force to the middle layer.

Jacquard weaving and weft-patterning proved suitable for localising complex functional structures to specific regions within wider textile areas while sealing the functional materials inside the textile. Jacquard weaving possesses particular potential for creating figurative elements: because Jacquard looms are widely used in the textile industry, the ability to support functional weaves of any size and shape within the limits of the loom setup is especially advantageous. This exploration's findings should be readily extendable to commercial textile manufacture.

The final exploration suggests that functional elements buried in the textile can be designed either to blend into the base fabric or be visually and texturally distinguishable from the surrounding area. Another critical finding is that the size of the piezoresistive sensor influences its measurement range. The shapes and motifs of any figurative sensor elements that have to operate within a specific measurement range must be designed to take this into consideration. There are ample opportunities for research into preferable sizes, shapes and motifs, whereby practitioners could refer to a library of figurative pressure-sensitive structures with a predefined value range, for easy application in interactive Jacquard graphics design. Design of the figurative motifs could benefit also from usability heuristics addressing how given visual cues might inform interaction with touch-sensitive interfaces. Section 5.2.2. expands on the implications of these findings for the tactile domain.

Finally, we acknowledge that Exploration 7 considered three-layer sensor structures only. While aware that this is not the highest layer count feasible for localised sensors, we should point out the challenge of managing additional weft systems as their number grows. To address such caveats and potential limitations, future studies should examine practical implementation of higher layer counts in woven e-textile sensor structures.

Exploration 7 showed that additional yarn systems can implement electrical or sensory features for individual areas or sides of a textile to construct location-specific functional elements. The inlay technique affords fully sealing away these inner elements and protecting them against abrasion. Meanwhile, the findings from Exploration 4 suggest that yarn systems can be designed also to interleave or switch order across layers and that, in additional shifting, specific yarns connected to functional areas can be led from inner layers to the surface and vice versa in the course of weaving.

These findings point to suitability for creating signalling structures and for connecting functional elements via conductive yarn systems. Demonstrating the practicality of integrating various woven-in components [90] opens exciting possibilities for real-world use of multi-layer structures, extending as far as fully woven circuits for fabric. We envision full-fledged woven e-textile systems as networks of woven sensors and enveloped components interconnected by means of conductive weft and warp lines that cross through the single-layered base fabric. That perspective facilitates holistic design of system-level circuitry [42, 63], force sensors [15] and integrated flexible filament circuits [50] all within the same woven e-textile. Attending to composite- and layout-level design in parallel can enable realising these structures in such forms as systems whose internal layer comprises non-linked pockets with the desired electrical connections for embedded hardware components, situated between the surrounding layers’ stable, durable linked structures. This is represented in portion 2 of Figure 17. Simultaneously, electrical signalling may be passed to designated locations at the textile's surface for interaction sensing purposes, as shown in portion 1 of the figure. In addition, surface signalling can serve to connect rigid printed circuit boards [58] or components and power sources [2] to the circuit. While some work on woven e-textiles has looked into mechanisms for interconnecting hard components and soft conductive textile materials [41, 56], their deployment within the inner layers requires further study.

In pointing to the value of our findings at composite and layout level, we wish to stress the importance of considering the two levels mutually in e-textile design. Firstly, designing them together facilitates a predictable woven structure as a solid environment for circuit integration in end applications. Since the density of a single-layer base fabric hinges on the weave structure and the weft-yarn thickness (as Exploration 7 clarified), practitioners must account for the density effects of their weave-pattern and material choices, which have knock-on effects on the electrical behaviour of the multi-layer composite. Likewise, designers adopting a broader perspective could take full advantage of the numerous options for the surrounding single-layer weave structure's construction to regulate the electrical performance of functional weaves. Further research in this area is clearly warranted.

Secondly, any interactive functional properties integrated into a textile substrate need to be created during the weaving. Since they cannot be embedded later on, production requires designing the circuit layout via weaving patterns. For example, the layout for wearable applications’ functional elements must be designed in conjunction with the sewing patterns for the final garment. This is entirely within the realm of possibility.

Explorations 5–7 attest that the surface characteristics and the electrically functional structures can be designed relatively independently. Ways of conveying the garment's interactive areas to the user can exploit this advantage. For visual patterns (stripes, shapes, texture variations created by the yarn types and weave patterns and so forth) consistent with the pattern of the embedded sensor arrangement, designers can choose weave patterns and materials for the surface that communicate messages about possible interaction, such as pushing or sliding gestures (3 and 4 in Figure 17). Reciprocally, the sensor structure may be designed explicitly to detect the intended forms of interaction. For example, the pressure-sensor structure proposed in Exploration 7 can be made to protrude from the fabric's surface by merging with 3D woven relief patterns [109] to afford pressing. A hairy tuft texture or long mohair fibres extending from the surface can act as an affordance for sliding gestures [26], in conjunction with an inner structure containing a capacitive grid [87] or slider [112] capable of detecting users’ gestures. These scenarios exemplify how a solid understanding of material, weave and pattern choices’ combined influence (e.g., on gestural interaction) permits designing the surface and the internal structures together.

Design principles and conventions for woven-textile-based interfaces are still in their infancy. The first steps toward a coherent gesture language and visual cues for e-textile interaction have been made, though [64]. By presenting our findings related to the versatility of multi-layer structures and surface patterning, we are taking the initiative to increase designers’ awareness of the opportunities and constraints as they design those interactions.

Developing tactile interfaces also calls for considering the experiential qualities of textile materials. Interfaces and applications anchored in day-to-day life [87] become coupled with the wearer's personal style and expression of identity [23, 24]. Hence, they call for discreet integration and ‘textileness’ [17]. Also, tactile experiences with textiles naturally evoke affective and emotional responses, associations and preferences [113, 114]. Therefore, designing a textile's look and feel for positive emotional responses requires not only knowledge of materials’ physical properties but also keen awareness of the psychological dimension of tactile perception [5]. The multi-layer weaving patterns presented in this paper free the interactive fabric's surface for expressive design—letting practitioners focus on materials, textures, colour combinations and figurative patterns that elicit positive responses. This freedom is essential for designing meaningful ways of interacting with technology through e-textile interfaces.