My paper “Heterotopia and Other Places: Displacing Expectations of Style and Theme in Lawrence Durrell’s Travel Books” is part of a new collection on Durrell, Heresy, and Heterotopia. It’s a great volume and I’m so excited to have it in hand to dive into the great work people are doing!
Since I have this digital space, I thought I’d use it to do a postmortem on the piece in print—and to provide a necessary correction. Specifically, I can’t help focusing on the figures that took me so long to adjust. Because I knew everything would be printed in black and white, I made sure to double code color alongside some other aesthetic like linetype or shape or position (or shape and position). Similarly, since I knew the dimensions to be used in the final printing, I could target output for legibility. These adjustments certainly helped! But I have room for improvement.
Figure 1 – Wrong trousers
For instance, here’s my first print-ready figure, in color as I saw it while polishing the piece, in the colorless gray scale I submitted for publication, and a photo of how it appeared in print:
Each version shows a comparison between three genres of Durrell’s prose: his travel writing (the main focus of my study), his literary fiction, and his commercial fiction. Comparing word frequencies on two axis, the figure demonstrates the possibly unsurprising relationship of genre to word choice. Travel writing is more likely to engage with the past by using the nouns “history” or “histories” to describe a shared, public timeline—historical buildings, infrastructures, regimes, and so on. Meanwhile, Durrell’s literary fiction relies more on the nouns “past” or “pasts” to describe an individualistic, personal experience of recently lived time. With notable exceptions, his commercial fiction is less likely to engage with either.
The color figure is most successful because of the details it gets right. The Okabe Ito palette ensures legibility for different abilities to distinguish color. Genre is double coded with both color and shape. And text for the travel writing genre is presented in black, letting it stand out as the main focus.
The grayscale version mostly hits the mark, but it sacrifices some legibility. When overlaying text layers, color was helpful to differentiate labels, but grayscale makes it all a jumble. And the grayscale gradations used for the two categories of fiction are hard to distinguish here.
The published figure is worse still. This version of the figure, using solid labels instead of a combination of shape and text, is from my initial version of the text I submitted for review,1 before any polishing for print. It relies only on color to differentiate, and overlaid labels fully block anything behind them. The spectrum of shades is also more limited than I realized: blacks look great, but grays are dithered from a pattern of small dots, leading to a loss of crispness. Lesson learned!
1 My paper is the only one in the collection using these kinds of figures as visual aid, so it’s very understandable that something might go a bit odd in the process. Still, it would be helpful to have had a proof stage where authors could review layout and catch minor errors before things are published.
Figure 2 – Genre distinctions
The second figure shows something that I still find fascinating. Partitioning texts by percentage and seeing what kinds of words are used shows that Durrell’s travel books tend to talk about traveling more in their beginnings and ends; his fiction shows no real pattern for where traveling is more likely to be found.
Color provides continuity between figures 1 and 2, but it wasn’t necessary, and the printed version looks similar to the grayscale version shown here; text sizing is slightly off because the figure was shrunk down from a larger version.
Figure 3 – Tangled words
Since the travel books and the literary fiction more readily engage with some idea of the past, figure three looks more closely at all the words they use. Words are ranked by how commonly they’re used in each group of books before being placed horizontally by their rareness in travel books and vertically by their rareness in Durrell’s literary fiction. Most words are overprinted in an undifferentiated mass—but some rise above, demonstrating words that are more common in travel writing than in the literary fiction. Trend lines show projected relationships of word rankings between the two genres: the solid line shows the actual relationship, where words that are pretty common in the travel books seem to be used more rarely in the novels; the dashed line shows an imagined relationship for comparison, showing how things would look if words were ranked equally in both genres.
As before, the grayscale figure sacrifices nothing in legibility, compared to the figure in color. The printed version is only mildly different from the polished version shown here, but the slopes do miss out on the increased legibility from the white borders.
Missing paragraph!
Unfortunately, a paragraph seems to have been dropped in the revisions process. The two paragraphs spanning from the bottom of page 226 to the middle of page 227 are missing their missing link, which should come in the middle. I’m including all three here, with the missing paragraph added in bold:
This figure [Figure 3] might seem to show evidence that Durrell’s travel books use rarified diction, surpassing even the literary works, but it actually shows the opposite. That the solid line rises more steeply demonstrates the travel books’ greater reliance on words common to both genres. Put another way, the literary works draw from a larger vocabulary, many words of which have been excluded here. Few words become legible by sinking below the tangle, and none that rise above it is surprising, considering that this latter group shows words used more often in books introducing islands that are Greek, featuring villages in valleys between mountains.
While this figure compares the words that are common among Durrell’s literary novels and his travel books, it does so by comparing differently sized genres. Six travel books are set against twelve literary novels. A better, balanced approach would approximate a reader’s experience, looking at smaller portions of text and comparing them to equally-sized sections from each genre. Figure 4 follows this approach. Each dot in this chart represents a single sample of 1,000 words taken randomly from a genre. The number of unique words in the sample is counted, and the dot gets placed horizontally based on this number. For each genre, the process is repeated 100 times, and the dots stack when more than one sample has the same number of unique words. Behind the dots, a shadow indicates a similar measurement with a density curve. Beneath them, box plots show the same data using a common statistical visualization: a middle vertical line shows the median value, a box emphasizes the distribution of the middle 50 values, and horizontal lines extend to show where remaining values fall; in this implementation, statistical outliers are ignored.
Despite Steiner’s and Kelly’s emphasis on precision and rarity of vocabulary in the literary novels, this process shows that Durrell’s travel books tend to have greater lexical diversity than his literary and commercial fiction. They use more words. Obviously, genres in this visualization show some overlap in their distributions. The figure even shows about half of the samples in the travel books seeming as if they could belong to the piles from either genre of fiction. But the other half of samples in the travel writing share diversity measures with the fiction more rarely, and the data suggest that the genres are truly different.
That missing second paragraph is crucial! Its introduction to Figure 4 sets up the third paragraph. Without the transition, the third paragraph seems to be devoted to Figure 3, describing it in ways that directly contradict the first paragraph. So I’m pretty sad the middle paragraph was dropped.
Figure 4 – Better comparison
As described in the paragraphs above, Figure 4 showcases a better method for distinguishing how often rare words are used in each genre. Equal-sized samples allow for direct comparisons, showing that the travel books are far less likely to repeat words.
This figure maintains continuity with previous figures by keeping colors and shapes consistent. But color isn’t necessary, so the grayscale version loses nothing for legibility. The printed version looks mostly the same, albeit with compromised scaling and no differentiation of shape.
Figure 6 – Final word
Skipping ahead, Figure 6 most strongly refutes the suggestion that Durrell’s travel books—written with great concern for financial stability but appreciated for their literary qualities—might fall stylistically somewhere between the other two prose genres. It shows how part-of-speech trigrams distinguish the works, with prose works separating most strongly by fiction and nonfiction. While the texts have more than 700 trigram patterns, I used principal components analysis to reduce them to two dimensions. The chart was created by sampling each genre, measuring grammatical patterns, and plotting them in this reduced space. Whole texts were then projected into the same space using patterns learned from the samples (by multiplying whole-text frequencies by the eigenvectors of the sample-based analysis).
Again, the printed version differs mildly from the grayscale because it uses a larger image shrunk down to fit the space, making the shapes harder to distinguish. Luckily, it doesn’t lose much for legibility.
Conclusions
At the end of the day, I’m very proud of the piece, and I’m proud to share what kinds of things it shows. I’m amazed, too, at all of the great works in the collection! As I finalize my next publication, I’m already considering the lessons learned here for best practices to adopt when preparing figures for publication—and I’m going to ask for page proofs to review before publication.
Citation
@misc{clawson2025,
author = {Clawson, James},
title = {New {Publication:} {Heterotopia} and {Other} {Places}},
date = {2025-09-23},
url = {https://jmclawson.net/posts/2025-09-grayscale-figures/},
langid = {en}
}