Claudia Berger is our Virtual Artist-in-Residence 2024-2025; register for their April 15th virtual talk and a local viewing of their data quilt in the Scholars’ Lab Common Room!

This year I am the Scholars’ Lab’s Virtual Artist-in-Residence, and I’m working on a data quilt about the Appalachian Trail. I spent most of last semester doing the background research for the quilt and this semester I get to actually start working on the quilt itself! Was this the best division of the project, maybe not. But it is what I could do, and I am doing everything I can to get my quilt to the Lab by the event in April. I do work best with a deadline, so let’s see how it goes. I will be documenting the major steps in this project here on the blog.

Data or Design first?

This is often my biggest question, where do I even start? I can’t start the design until I know what data I have. But I also don’t know how much data I need until I do the design. It is really easy to get trapped in this stage, which may be why I didn’t start actively working on this part of the project until January. It can be daunting.

N.B. For some making projects this may not apply because the project might be about a particular dataset or a particular design. I started with a question though, and needed to figure out both.

However, like many things in life, it is a false binary. You don’t have to fully get one settled before tackling the other, go figure. I came up with a design concept, a quilt made up of nine equally sized blocks in a 3x3 grid. Then I just needed to find enough data to go into nine visualizations. I made a list of the major themes I was drawn to in my research and went about finding some data that could fall into these categories.

What my initial planning looks like.

But what about the narrative?

So I got some data. It wasn’t necessarily nine datasets for each of the quilt blocks but it was enough to get started. I figured I could get started on the design and then see how much more I needed, especially since some of my themes were hard to quantify in data. But as I started thinking about the layout of the quilt itself I realized I didn’t know how I wanted people to “read” the quilt.

Would it be left to right and top down like how we read text (in English)?

Or in a more boustrophedon style, like how a river flows in a continuous line?

Or should I make it so it can be read in any order and so the narrative makes sense with all of its surrounding blocks? But that would make it hard to have a companion zine that was similarly free-flowing.

So instead, I started to think more about quilts and ways narrative could lend itself to some traditional layouts. I played with the idea of making a large log cabin quilt. Log cabin patterns create a sort of spiral, they are built starting with the center with pieces added to the outside. This is a pattern I’ve used in knitting and sewing before, but not in data physicalizations.

A template for making a log cabin quilt block by Nido Quilters

What I liked most about this idea is it has a set starting point in the center, and as the blocks continue around the spiral they get larger. Narratively this let me start with a simpler “seed” of the topic and keep expanding to more nuanced visualizations that needed more space to be fully realized. The narrative gets to build in a more natural way.

So while I had spent time fretting about starting with either data/the design of the visualizations, what I really needed to think through first was what is the story I am trying to tell? And how can I make the affordances of quilt design work with my narrative goals?

I make data physicalizations because it prioritizes narrative and interpretation more than the “truth” of the data, and I had lost that as I got bogged down in the details. For me, narrative is first. And I use the data and the design to support the narrative.

Time to sketch it out

This is my absolute favorite part of the whole process. I get to play with dot grid paper and all my markers, what’s not to love? Granted, I am a stationery addict at heart. So I really do look for any excuse to use all of the fun materials I have. But this is the step where I feel like I get to “play” the most. While I love sewing, once I get there I already have the design pretty settled. I am mostly following my own instructions. This is where I get to make decisions and be creative with how I approach the visualizations.

(I really find dot grid paper to be the best material to use at this stage. It gives you a structure to work with that ensures things are even, but it isn’t as dominating on a page as a full grid paper. Of course, this is just my opinion, and I love nothing more than doodling geometric patterns on dot grid paper. But using it really helps me translate dimensions to fabric and I can do my “measuring” here. For this project I am envisioning a 3 square foot quilt. The inner block. Block 1, is 12 x 12 inches, so each grid represents 3 inches.)

There is no one set way with how to approach this, this is just a documentation of how I like to do it. If this doesn’t resonate with how you like to think about your projects that is fine! Do it your own way. But I design the way I write, which is to say extremely linearly. I am not someone who can write by jumping around a document. I like to know the flow so I start in the beginning and work my way to the end.

Ultimately, for quilt design, my process looks like this:

1. Pick the block I am working on
2. Pick which of the data I have gathered is a good fit for the topic
3. Think about what is the most interesting part of the data, if I could only say one thing what would that be?
4. Are there any quilting techniques that would lend itself to the nature of the data or the topic? For example: applique, English Paper Piecing, half square triangles, or traditional quilt block designs, etc.
5. Once I have the primary point designed, are there other parts of the data that work well narratively? And is there a design way to layer it?

For example, this block on the demographics of people who complete thru-hikes of the trail using annual surveys since 2016. (Since they didn’t do the survey 2020 - and it was the center of the grid - I made that one an average of all of the reported years using a different color to differentiate it.)

I used the idea of the nine-patch block as my starting point, although I adapted it to be a base grid of 16 (4x4) patches to better fit with the dimensions of the visualization. I used the nine-patch idea to show the percentage of the gender (white being men and green being all other answers - such as women, nonbinary, etc). If it was a 50-50 split, 8 of the patches in each grid should be white, but that is never the case. I liked using the grid because it is easy to count the patches in each one, and by trying to make symmetrical or repetitive designs it is more obvious where it isn’t balanced.

But I also wanted to include the data on the reported race of thru-hikers. The challenge here is that it is a completely different scale. While the gender split on average is 60-40, the average percentage of non-white hikers is 6.26%. In order to not confuse the two, I decided to use a different technique to display the data, relying on stitching instead of fabric. I felt this let me use two different scales at the same time, that are related but different. I could still play with the grid to make it easy to count, and used one full line of stitching to represent 1%. Then I could easily round the data to the nearest .25% using the grid as a guide. So the more lines in each section, the more non-white thru-hikers there were.

My last step, once I have completed a draft of the design, is to ask myself, “is this too chart-y?” It is really hard sometimes to avoid the temptation to essentially make a bar chart in fabric, so I like to challenge myself to see if there is a way I can move away from more traditional chart styles. Now, one of my blocks is essentially a bar chart, but since it was the only one and it really successfully highlighted the point I was making I decided to keep it.

These are not the final colors that I will be using. They will probably all be changed once I dye the fabric and know what I am working with.

Next steps

Now, the design isn’t final. Choosing colors is a big part of the look of the quilt, so my next step is dyeing my fabric! I am hoping to have a blogpost about the process of dyeing raw silk with plant-based dyes by the end of February. (I need deadlines, this will force me to get that done…) Once I have all of those colors I can return to the design and decide which colors will go where. More on that later. In the meantime let me know if you have any questions about this process! Happy to do a follow-up post as needed.

Two months into this fellowship, I have prayed in the following places:

• The Grad lounge
• Brandon’s office
• Shane’s office
• Amanda’s office

The first time, it felt strange. I had barely known everyone for a week. I didn’t want to make anyone uncomfortable. I didn’t want to seem like I was putting on a show of religiosity. I didn’t want to be stereotyped and put into a box.

Each time I asked if I could pray in the Scholars’ Lab space, those around me were extremely accommodating, offering to leave the room to give me privacy. That made it feel like even more of an imposition. I felt too conspicuous, too seen. The kinder everyone was, the more uncomfortable I felt. I couldn’t make sense of it. Why did this kindness make me feel like an outsider?

Soon enough, the afternoon prayer started eliciting other uncomfortable thoughts. Once, as I unfurled my prayer mat, I wondered if the DH tools we discovered would ever support Punjabi or Urdu (my research languages). Shane and I had spent an entire morning trying Tesseract’s OCR software on images with Persian, Urdu, and Punjabi text, but the invariable result was gibberish. A few weeks later, when I wanted my name in both English and Urdu on our Charter website, Jeremy said he’d figure out if and how that was possible. I nearly told him to forget I mentioned it. I remember noticing how brown my skin was as I prayed that day.

The experience of double consciousness each time I pray in the Scholars’ Lab is a stark reminder that I don’t fully belong in the ‘Digital’ Humanities. I have to be accommodated for, adjusted to, and worked around. It doesn’t matter how sincerely the Scholars’ Lab staff welcome me into their physical space. As soon as we face a laptop screen, I am stripped down to an anglicized, areligious, apolitical version of myself. For the computer only recognizes these fragments. Here, too, it has become the job of the SLab folks to stretch themselves in unexpected ways to make me whole again: by trying to find digital platforms and tools with Right-To-Left (RTL) language support; by hunting down essays on Global DH and Minimal Computing; by dredging up their own insecurities and limitations in conversations to assure me of my place in DH.

The message is clear: It takes the kindness and effort of individual DH scholars to make space for me within systems that were not designed for people like me. Grateful as I am, it is not kindness I want, but the chance to be an equal collaborator. To create and share knowledge across the linguistic communities I belong to.

In a recent paper, Masoud Ghorbaninejad, Nathan P. Gibson and David Joseph Wrisley have discussed the Anglocentric nature of current DH infrastructures that largely ignore the “digital habitus”¹ of RTL language users. They state that “knowledge is not just cultural content embedded in language; it is also infrastructure that allows that content to be represented, circulated, and preserved for the concerned communities.” Of the many tools I have discovered these past few months – Omeka, Voyant tools, MALLET, Tesseract, to name a few – not a single one supports Urdu or Punjabi in any meaningful way. As a multilingual South Asian and a student of Muslim literatures, each interaction with these tools involves two things: (1) silencing the very voices within me that have already undergone violence at the hands of the English language, and (2) a fervent hope for alternatives.

(Thank you Brandon for the title!)

A Co-authored Series of Posts ‘About 1919,’ that is, about English-language books published from 1914 to 1921, according to the online bibliography and database, Collective Biographies of Women.

Collective Biographies of Women’s eponymous genre almost naturally gives rise to network analysis. In the same way that researchers today look at social networks (online or otherwise) and think about the meanings of “connections” or “nodes” within them, so too can we look at collective biography as a mode of textual and social connection. Collective biography is sometimes called prosopography, a method of studying sets of people as “parallel lives,” to echo the title of Plutarch’s classic prosopography of the lives of Greeks and Romans (widely read in the nineteenth century). Each time we look at a table of contents in CBW, we are presented with a suggested grouping. Here, the author and publisher say, are women who belong together. Yet we call these connections a documentary social network, unlike an actual one built by letters, meetings, relationships. The subjects of the chapters in a collection often had no interaction with each other but lived in different centuries, countries, or social circles. Nevertheless, prosopography strives for measurable comparison by carefully documenting as much as can be known about individuals in the comparative network.

A woman’s life narrative may belong in more than one place. One of the main ways researchers at UVA have been using the CBW database has been to compare the recognition rate, our term for how many times an individual appears in tables of contents across the database. This rough measure of person A’s and person B’s relative usefulness as examples can be even more revealing in our study of networks, discussed in a moment: how often do A and B get placed together in a book of women’s lives? For now, think of a woman like Abigail Adams, who predictably appears in collections of revolutionary women, of first ladies, of the mothers of presidents, and of American women more generally. See: the prolific historian Elizabeth Fries Lummis Ellet’s The Women of the American Revolution, first published 1848 a259; almost a century later, Kathleen Prindiville’s First Ladies in 1932 a654. During World War I, Adams resurfaces in William Judson Hampton’s 1918 collection, Our Presidents and Their Mothers ([a374] (http://cbw.iath.virginia.edu/books_display.php?id=1705); and again in Gamaliel Bradford’s Portraits of American Women in 1919 (a102). Bradford’s distinctive career as a biographer is featured in Mackenzie Daly’s post, third in this series.

What might it mean for a person–versions of a historic woman–to occupy various positions, inherent in the tables of contents she appears in? What kinds of documentary social networks arise? In the realm of collective biography, how can we quantifiably discuss “connectivity”? Probably, a high “recognition rate” (RR) will indicate that a person was (at the times when the collections were published) far from obscure. But this measure of status may not correlate with the actual rank or power of a particular woman during her life. We see some isolated, single-biography subjects in CBW were once upon a time queens. Some 6000 women appear only once in CBW’s 15,000+ biographical chapters. All of these measures resonate with the narratives themselves as our team tags them with an XML schema. Does a person’s relative connectivity correspond to that murkier (but perhaps more intuitive) notion of “obscurity”? Do either of these aspects of a person make any difference to how their story is told?

In this blog, I’ll try to answer each of these questions in turn. As examples we’ll look at two collections both published 1919, just in the aftermath of World War I. The first was noted above for including Abigail Adams: a102 Portraits of American Women by Gamaliel Bradford (Houghton Mifflin, Boston). Its counterpart here is a108 Women of Achievement by Benjamin Griffith Brawley (Women’s American Baptist Home Mission Society, Chicago). These volumes make different claims to speak for the mainstream, by highly educated men, one an academic historian, addressing general readers.

Table of Contents for a102 Portraits of American Women (HathiTrust)

Table of Contents for a108 Women of Achievement (HathiTrust)

As the word “Negro” in the table of contents underlines, Brawley’s featured women have achieved in a context of racial discrimination. Readers might have guessed a criterion of selection that the titles don’t tell by looking at the publishers and cities on the copyright pages: Houghton Mifflin is a prominent New England publisher to this day and the American Baptist Home Mission and its Women’s offshoots in the East and Midwest continue to work with the poor and people of color. A closer look, and we find that all of Bradford’s eight subjects are white, without remark. Brawley’s five biographical chapters portray African-American women under the non-racial rubric, “women of achievement.”

“American” has a meaning in both books that presumes whiteness as default. “Home” refers to missions like the Salvation Army (originating in London as “Christian Mission” and launched in the U.S. in 1880) or Jane Addams’ Hull House that serve onshore rather than in foreign lands. Addams, 1860-1935, RR=21, and many other American female leaders of home missions were white, not Baptist, and would not fit in Brawley’s book. Collective biographies published in the U.S. often focus on varied nationalities, yet these two are invested in defining American women.

The database and other resources identify Bradford as a white man and Brawley as a Black man, contemporaries. Gamaliel Bradford VI, born in Boston before the US Civil War (9 October 1863 – 11 April 1932), was the grandson of an abolitionist namesake and educated at Harvard, who became a well-known biographer (see Mackenzie Daly’s blog). Brawley (22 April 1882 - 1 February 1939) was born in a middle-class minister’s family in South Carolina and earned a BA from University of Chicago and MA from Harvard before serving as Dean of Morehouse College and later teaching at Howard; he was a prolific poet and published a range of histories of African Americans, college texts and literary criticism, and autobiography. Brawley in his day was well-known in African-American elite circles. Bradford gets mentioned in literary histories of biography.

These rather short-list collections identify a range of female achievers most of whom are well-known now, though early in the twentieth century biographical sources on the recent and living Black women would have been scarce. Whereas Bradford’s “psychographies” portray a New England “American Renaissance” canon before the twentieth century, Brawley’s starring roles in 1919 include Meta Warrick Fuller, a versatile artist who lived until 1968, and national leaders Mary McLeod Bethune and Mary Church Terrell, whose most famous achievements occurred between World War I and their deaths in the 1950s. Both biographers portray female character and public action, broadly a cause of progress for their sex and their nation, while Brawley’s volume overtly supports an organized social mission. There is a race-based time lag, in a sense; activism calls for recognition of more obscure, living workers for the cause.

Documentary Social Networks and Degrees of Separation

As with other social networks, these women can be thought about in terms of degrees of separation. To be precise—if Woman 1 and Woman 2 appear in the same book, they have a degree of separation designated 1 (the CBW team has dubbed them “siblings”). If Woman 2 and Woman 3 appear in the same book, but Woman 1 and Woman 3 do not appear in the same book, then Women 1 and 3 have degree of separation 2 (and we can call them “cousins”). In this blog post, we shall consider as many degrees of separation as possible for a woman: call them, perhaps, third or fourth cousins. While we see the significance of this sort of measure by looking at two 1919 publications, I used the searches of tables of contents across the database of 1272 books to generate relative degrees of obscurity or connectivity. (CBW’s radial graph for visualizing degrees across tables of contents is being redeveloped; ask Alison Booth, email ab6j@virginia.edu.)

For each woman in the two collections I looked at, I determined their connectivity within our database of collections by finding the total number of women listed in contents at each degree of separation, up to 6 degrees of separation. I then used those numbers to determine “average degree of separation”—let us, to use a spicier term, call it “d-factor.”

Thus, Abigail Adams has 601 siblings, 3994 cousins, and so on. Here are the results for Portraits and Achievement represented in tabular form:

A102 Bradford, Portraits

Person	1	2	3	4	5	6	d-factor
ADAMS	601	3994	2783	148	0	0	2.32925857
RIPLEY	7	1416	5600	487	16	0	2.878952963
LYON	201	4893	2347	85	0	0	2.307733192
STOWE	588	6218	691	29	0	0	2.021392506
FULLER	418	5700	1340	68	0	0	2.140579325
ALCOTT	467	6105	903	51	0	0	2.071485517
WILLARD	224	5375	1875	52	0	0	2.233191602
DICKINSON	180	4766	2467	113	0	0	2.333909115

A108 Brawley, Achievement

Person	1	2	3	4	5	6	d-factor
TUBMAN	482	471	2244	85	0	0	2.256675967
GORDON	187	580	5281	1422	57	0	3.077321642
FULLER	30	797	5245	1420	35	0	3.08409725
BETHUNE	341	4786	2284	116	0	0	2.288959745
TERRELL	188	579	5281	1422	57	0	3.077188787

Stowe and Alcott have the lowest d-factors of this array of women, that is, we might say the highest connectivity. And here is the data presented graphically. In the graphs below, the X axis represents degrees of separation and the Y axis represents the number of connections at that degree of separation.

A102 Bradford, Portraits

A108 Brawley, Achievement

It’s difficult to assign too much meaning to the absolute values of these women’s d-factors, but these measures add significance in comparisons and contexts across CBW, including the narrative methods themselves. Further analysis might show tendencies such as fewer episodes or descriptive passages correlating with longer tables of contents and perhaps higher rates of “singleton” subjects (RR=1). To write a biography of a rare subject or a living person who has been little documented generally takes more time and effort than to patch together and embroider from previous printed sources. For the audience, too, the familiar biographical subject may be preferable. The supply and demand economics seem not to apply strictly when circulating the name and narrative of a historic person. Less separation (closer association between two women and their associates in turn) can indicate being closer to the centrality of a national history, and higher probability of cropping up in a table of contents of lives of women of any type. Ripley, with RR=1 (solely in Bradford) has a d-factor slightly larger than Mary Lyon, RR=20.

It’s more interesting to think about them in relation to each other. We can say that a woman with a d-factor of 2 is certainly more closely linked to the mass of other women appearing in our database than a woman with a d-factor of 3. A woman with a d-factor of 2 is more closely connected to the rest of the database than a woman with a d-factor of 2.3; though this difference is nowhere near as intense as the first example, it is still quite significant. In the tabular data above, for instance, look at the disparity between Louisa May Alcott and Harriet Tubman. Though both are 1 degree of separation from roughly the same amount of women (467 vs. 482, respectively), Alcott is 2 degrees of separation from over 1000 more women than Tubman is. This difference, which could be attributed both to the wider social networks of a white woman in New England who published widely and to the bias of CBW books’ documentary networks–their tendency to privilege white women writers, for example. Their respective d-factors, in small but telling averages, reflect this: Alcott 2.07; Tubman 2.25.

You might expect d-factor to simply correlate to the number of chapters about a woman in the volumes in the database, her recognition rate (RR) as we call it. But at a different level d-factor helps showcase how some women are more likely to be connected to specific well-connected women. It is another way of displaying “fame” or “notability,” and with d-factor we can begin to surmise about the distinctions in obscurity between women brought together into the same collection. Some, I suppose, will be “central nodes,” and others will be more “peripheral.”

As we can see from both representations (but more quickly in the graphs), both Bradford’s and Brawley’s collections have two identifiable subsets of women. Broadly, Portraits is divided into Sarah Alden Ripley and everybody else, with Ripley the least “connected” woman with RR=1 (Bradford alone of 934 authors of CBW’s texts decided to include her in a collection). Comparably, Achievement is divided between Gordon/Fuller/Terrell and Tubman/Bethune as, respectively, less connected and more connected women. An effort has been made to equilibrate the two graphs’ size, but note that their y-axes have different maxima.

Achievement’s women seem to have low deviance within those two groups, at least compared to the group of Portraits’s women with a d-factor near 2. Particular attention ought to be paid, again, to Abigail Adams, who in spite of having the highest number of siblings (1 degree from her), has a relatively low number of people 2 degrees of separation from her. We speculate that this lower indirect (cousinly) connectivity may be due to Adams’s tendency to appear in the same book with wives or mothers of U.S. presidents, as we already have seen, rather than in more mixed tables of contents with varied subjects. We could explore this for other types of American women.

Now, does any of this affect the way the author tells the story? That is, does a person’s relatively low d-factor (a quantity that the author would not directly know but may have some proxy conception of)—change how their story is told? Let us localize this discussion to the introductions of the works in question.

We see an authorial awareness of obscurity in the first paragraph of Bradford’s lone obscure biography, for Sarah Alden Ripley (who happened to be his relative): Few American women of to-day know of Mrs. Samuel Ripley, but a sentence from Senator Hoar’s “Autobiography” will give her a favorable introduction: “She was one of the most wonderful scholars of her time, or indeed of any time. President Everett said she could fill any professor’s chair at Harvard.” Contrast this introduction with the way he introduces Frances Willard (“She had the great West behind her; its sky and its distances…”), Louisa May Alcott (“Her father thought himself a philosopher…”), and Harriet Beecher Stowe (“She was a little woman, rather plain than beautiful…”). With Ripley there is something of a necessary justification, an initial pivot to a secondary source which proves that she belongs with the subjects of the other chapters, and at the same time an implicit criterion of inherent, merited but uncrowned achievement. With Ripley as contrast, we see more clearly what the biographer can do when they work with a well-known biographical subject: dramatize, begin in medias res, take advantage of the fact that the reader is coming back to someone familiar, so that the introduction is really a reacquaintance. Ripley, with a higher d-factor, earns Bradford’s least artful flourish of narrative technique in her opening sentence: the lower the d-factor, the less an introduction needs to actually introduce.

That, at least, is the conclusion I can draw from Portraits. But does looking at Achievement change that thesis at all? Let us see how Brawley introduces his two “well-connected” women (Tubman and Bethune) vs. his three “obscure” women. [^bignote]

Tubman: “Greatest of all the heroines of anti-slavery was Harriet Tubman.” Bethune: “On October 3, 1904, a lone woman, inspired by the desire to do something for the needy ones of her race and state, began at Daytona, Florida, a training school for Negro girls.”

Gordon: “This is the story of a young woman who had not more than ordinary advantages, but who in our own day by her love for Christ and her zeal in his service was swept from her heroic labor into martyrdom.” Fuller: “The state of Massachusetts has always been famous for its history and literature, and especially rich in tradition is the region around Boston.” Terrell: “With the increasingly complex problems of American civilization, woman is being called on in ways before undreamed of to bear a share in great public burdens.”

It is more difficult to make the same argument out of these quotes as we did with the Bradford book. Might Brawley be working in a different mode altogether? We see that Tubman, by far his most famous subject, least requires citation to an external authority in his introduction. Alternatively, we might say that Tubman’s importance demands his respect, his summative statement of her greatness. Could there be a difference to writing Black biographies vs. writing white biographies, that manifests in this particular manner?—that is, to deal with a “well-known” persona unlocks a white biographer’s capacity to do away with formality, but absolutely necessitates formal announcement for a Black subject. Bradford indulges in his breezy approach favoring interiority and impressions upon the security of generations of New England insider status. Brawley, in spite of his academic credentials and prestige, knows his Black women needed to dress in their Sunday best to be classified as “American women” or “women of achievement” without the racial modifier. Such inequities might appear in the more formal, grander dressing of Brawley’s subjects.

These are just a few considerations, a starting set of theses. As with much digital humanities research, this brief discussion has offered more questions than it has answered. It has hoped to provide a new quantity for potential analysis, applicable most readily in the CBW database but, with some trivial modifications, easily applicable in similar databases representing networks of people–similar prosopographies.