Knot Hard: Accessible Textile Data Visualization with a Circular Knitting Machine

“That is such a cool idea, but I’m not crafty at all.”

I hear this a lot when I mention that I run a Textile Makerspace, and teach Data Visualization with Textiles. It’s easy to see where people are coming from: temperature blankets, covid scarves, quilted topographic maps, embroidered Greek epic poetry — the data visualizations with textiles that have filtered into the attention of the general public all involve a significant amount of time and skill and thoughtful creativity. But the lure of being able to wear your data, or just turn it into a thing you can hold in your hands, draws in not only those who have cultivated a serious maker-hobby. This is where a circular knitting machine shines.

I learned to knit for the first time in high school, where I passed time in English and History classes making a small shoulder bag and a scarf: my only two creations before setting the craft aside. Even once I learned to sew and started making my own clothes, I remained certain that knitting wasn’t for me — and doubly so once I got a serger (a specialized sewing machine) to sew sweaters out of knitted fabric. I was too impatient for knitting, I told myself, overlooking the well of patience I’d managed to dig for myself between having three kids and learning to code. When a library colleague, Amanda Whitmire, started posting her adventures with a circular knitting machine, it felt like a way to open up new paths for data visualization without committing to learn a craft that seemed overwhelming in its time- and skill-demands. All you have to do is crank a handle, and yarn turns into a scarf.

The four vignettes in this piece trace my attempts to incorporate the circular knitting machine into my developing practice of textile data visualization, illustrating some of the ups and downs of learning to translate data into machine knitting. I’ll conclude with a few thoughts about why you might want to consider adding a circular knitting machine to your makerspace, or as a machine that you can share with others for team-building activities, or mental health breaks.

Why you might want a circular knitting machine

Circular knitting machines — particularly small ones like the 22-needle AddiExpress machine — are small, more affordable than most textile equipment (around $120), highly portable (thanks to the legs unscrewing), and a very accessible way to nudge people into textile making, even people who self-identify as “not crafty at all”. They’re fairly sturdy, but if someone uses a yarn too thick for the machine and pushes hard enough to break a needle, replacing a needle takes less than 10 minutes and is not a big deal, even if you’re not super comfortable disassembling machinery. If you can crank a handle, you can make a scarf. If you can count things and make a spreadsheet and do some very basic math in that spreadsheet, you can make a data scarf that can be both beautiful and filled with meaning — if you choose to explain it. By only having a small number of needles in a single gauge and fixed spacing between them, the circular knitting machine constrains the decision space of textile data visualization, in a way that is helpful for those new to this approach. Still, there are many facets you can play with: color, material (fiber content of the yarn), texture, length (of the piece overall and individual pieces of yarn), and patterns — none of which require textile craft skills beyond tying knots and cranking the handle. For those with other craft skills and interests– including embroidery, crochet, sewing, and hand-knitting– these can be fruitfully combined with knitted scarves or panels to add nuance and complexity to the textile data visualization.

Standing meetings

An internal grant funded a small and large Addi circular knitting machine for the Makerspace in January 2023, and I went looking for data to knit with this new tool. The most obvious use of it is to make scarves, which fit well with the season. I quickly settled on my own calendar: what were my standing meetings for the previous year? How often were they held vs. canceled? Like so many DH projects, it started with a spreadsheet: one row for each week, one column for each weekly standing meeting (my academic department, my immediate organization in the library, the Humanities and Area Studies librarian meeting I attend, and the Research Data Services group that had recently formed). One symbol if the meeting were held, another if it were canceled, and additional symbols if I was on vacation or a work-related trip during that meeting’s time slot.

I figured out that the yarn needed to knit one row on the smaller machine stretched roughly from my shoulder to fingertips, but it was far from precise. A great deal depended on how uniformly I tied the knot connecting the bits of yarn, and consistency is still not my strength. As one early experiment after another failed with dropped stitches leaving gaping holes, I concluded that tying the yarn together as I was knitting was making the task harder than it had to be. Stopping and starting the machine to tie knots was leading to dropped stitches much more often than if all the data were prepped in advance into a neat ball where I could crank the handle continuously and evenly. Thus the data ball was born, a physicalization of the kind of cleaning and reformatting work we so often do with data, transforming it from a spreadsheet into something knittable. The parallels with other DH projects were striking: the biggest time sink came from gathering the data, and preparing it in the right way. This work feels minimally-skilled, but in practice it involves moments of decision-making that can shape the outcome in ways that can be hard to anticipate in advance. Should consecutive black segments be cut and tied together, since they represent discrete cancellations? Or for the sake of expediency should multiple cancellations just be a single, long piece? And how to guesstimate the length, to account for the knots that would not need to be tied? 

After all the preparation, knitting a scarf– much like creating charts or graphs– took almost no time at all.

This first project continues to be one of my favorite examples of textile data visualization. By chance, around 200 rows turned out to be a pretty reasonable scarf length, though on future ones I’ve leaned more towards 250. The colors I picked and the frequency of canceled meetings made for a visually attractive striped scarf, even without an explanation of the data. It also serves as a good example of the subversive language of textiles, making a wearable that lays bare — if you know how to read it — the mismatch between a meeting’s stated importance and the number of times it was actually held.

Love Data Week

Like many libraries, Stanford Libraries organized a series of events for Love Data Week in mid-February 2023. The sequel to the standing meetings scarf was born from a fit of pique: I was irked at this event telling me how I should feel about data. “How,” I wondered, “do people actually feel about data? What if we went descriptive instead of prescriptive?” The morning of the Love Data Fair, I made several copies of a survey with a single question, asking people to share their feelings about data. I posted one to the library Slack, another to Mastodon (where I had been spending time online since Elon Musk’s Twitter takeover), and I asked a colleague to post the third to Twitter. 

I made a mistake with the survey I soon came to regret, allowing an “Other” option with a write-in, in addition to the options “I love data”, “Mixed feelings”, “I hate data”, “No feelings”. Qualitative data is more challenging to translate into yarn, especially when you’re only thinking in terms of yarn colors and a uniform mechanical stitch.

I spent the Love Data Fair preparing the data ball. This one was simpler than the standing meeting data scarf, since that scarf represented multiple facets: meetings (yarn colors) and time (scarf length), while this one would only represent survey responses. For each data source, I simply aggregated and counted the survey responses, which led to long sections of each color. As people came by the Textiles + Data booth at the Love Data Fair, I asked them about their feelings about data, adding their results to a smaller data ball that would eventually be joined to the main one. 

In the last half-hour of the fair, I clamped the circular knitting machine to a table and invited people to help with the knitting. This invitation was met with the same mix of puzzlement, curiosity, and wariness one often finds in the digital humanities classroom: Do I trust myself to not break this machine? What if I make a mistake? Is this actually a good idea? All my tech- and data-oriented colleagues faced with this bizarre new machine are very used to those reactions from students in workshops, but it’s another thing entirely to experience those anxieties yourself with a technology unfamiliar to you. To my surprise, textiles had become a tool for cultivating empathy.

Data-Sitters Club

Since 2019, The Data-Sitters Club has been collaborating on a friendly, colloquial guide to computational text analysis, using Ann M. Martin’s The Baby-Sitters Club series. We had never all met in person until March 2023, when Roopika Risam threw a Data-Sitters Club retreat over a couple snowy days at Dartmouth. Snacks and crafts were essential supplies for such an event.

Over the course of a couple days, and further inspired by a visit to Jacque Wernimont’s Digital Justice Lab (well-stocked with a digital knitting machine, looms, fibers of all sorts, and some marvelous creations in various stages of development), the Data-Sitters gathered their data, assembled data balls, and almost everyone went home with a scarf. It helped that we all knew each other and had plenty to talk about while constructing the data balls, but a similar setup could also work well for an open workshop: find and prep an interesting data set, offer some basic tools for analyzing it in different ways (e.g. Voyant if you’re working with a text data set), and let people come up with different questions about the data alone or in groups. Those groups could then work together to get the numbers they need, and each person could assemble and knit their own data ball.

Lee Skallerup Bessette worked with the Baby-Sitters Club Fandom Wiki to find out how many of the Baby-Sitters Club books were translated into each language identified there, and make one row per book per translated language. Anouk Lang and I worked with our own data about the Ann M. Martin Papers at Smith College, charting which book had which kind of supporting materials (e.g. synopses, outlines, other notes) in the archives. Maria Cecire and I calculated the average sentence length for each book, and she used that data. Katia Bowers focused on the publication history of the Baby-Sitters Club books, month by month, from 1986-2006.

Knitting the scarves took time, and spilled over into the sleepover portion of the retreat: not all the yarns were equally cooperative with the machines, and we had to go slowly and carefully to avoid dropping stitches. Dropped stitches can be recovered while the knitting is still on the machine using a crochet hook to connect it through however many rounds have passed since the stitch was missed; sometimes, though, it’s easier to just secure the stitch after the fact by passing a small length of yarn through the floating loop and tying it on the inside of the scarf, and I patched up a few mistakes on the scarves that way.

As the owner of the knitting machine, I saved my data knitting for the flight home. Here I can say that attaching the circular knitting machine to the tray table and using it on a plane is technically possible but not recommended. The jostling of the plane made it easier to drop stitches, and the long screws of the clamps poked uncomfortably– even as someone much smaller than an airplane seat. There were enough dropped stitches at the end that I unwound the scarf and swore I’d re-knit it, but the data ball remains in my office.

 

Hong Kong

In May 2023, I visited Hong Kong University as part of the New Horizons for Digital Humanities event organized by Javier Cha. While my official contribution to the event was a talk on digital cultural heritage preservation and a workshop with Melanie Walsh on multilingual NLP using fanfic data, I also wanted to bring in the textiles that were becoming an increasingly meaningful part of my work. I held a lunch workshop on data visualization with textiles, talking through a couple of example projects and the tools I’d used to make them. As a follow-up, I enlisted all the workshop participants — in-person and virtual — in a brief survey about how long they had traveled to attend: ranging from 0 minutes (virtual participants) to over 30 hours (Gimena del Rio Riande, whose route from Buenos Aires took her via Addis Ababa). 

I’d traveled with just my backpack and one carry-on, but packed it full of yarn, roving (wool for spinning into yarn), along with my travel e-spinner for creating the yarn, and a handful of beads. I consulted with Gimena on mapping the survey results to colors; this time I wasn’t trying to make a scarf, but just a flat-panel visualization. The circular knitting machine has a flat-panel switch that treats the black needles as stoppers — so you go back and forth between, rather than in full circles. It takes a little practice to get the hang of it: you have to turn the handle a little further than you think you should, as if to start a stitch on one of the black needles, but then that stitch is caught on the first white needle when you change directions.

The panel was only the first part of this visualization: I wanted to capture both time and space in people’s travel. I asked the student assistants at the conference to print me out an image with a Dymaxion projection of the world map (my favorite projection), and a map of Hong Kong, which I set about attempting to embroider in yarn on the knit panel. It was a frustrating and illuminating experience: the knitting machine does not produce very dense fabric, which limits the amount of detail you can include. If part of the goal of the Dymaxion projection is to disorient viewers used to a centered view of North America and Europe on their maps, an embroidered map on a knit panel is all the more disorienting through its inevitable distortion of land shapes. 

The Hong Kong embroidery had similar challenges, trying to include enough detail to be recognizable with prompting, but also ensure enough room to capture people’s travel paths. On top of it all, I struggle with geography even under ideal circumstances, so trying to translate from locations on Google Maps, to corresponding locations on my paper map cut-outs, to where that would be on the knit panel, was no small challenge.

I worked with my fellow panelists to trace our flights — not trying to mimic actual flight paths, but just getting a piece of string from a departure airport to its destination. Red beads marked the airports we’d left from and orange ones marked the airports we only transferred through.

For the local attendees, yarn color indicated travel mode. I used a darker blue for MTR (subway) transit, yellow was for bus, and pink was for car. Walking paths were a spun single (rather than plied) yarn strand because they were so short. One participant wanted to represent rather than simplify his complex data, where he uses one means of transit to drop off his daughter, and another to go from her school to the workshop. For that, we tied two colors of yarn together, and put the knot in the center of the path.

The final piece, given to the organizers as a thank-you gift, was a joyfully chaotic thing, made collaboratively and with love for the circumstances that brought us together for that week. 

Conclusion

When I got the circular knitting machine, only a bit over a year ago as I write this, I was fully convinced that this would be how I would interact with knitting and the yarn-based arts in general. A year later, I’ve picked up crochet and become very fond of hand-knitting, and the flexibility it offers in contrast to the machine. Even though I don’t use mine regularly anymore, I wouldn’t have ended up here if I hadn’t started with the circular knitting machine, and I’m quick to bring it out for workshops and events, in case it sparks that same journey for someone else … or just lets them create a really cozy scarf in less than a half hour.