A comment by a reader on a previous post (contrasting my Ottoman timechart with Minard’s famous map of Russia 1812) merits further discussion. Warning: theoretical discussion of the visual display of (not just) quantitative information follows.
Minard’s map is considered successful because it makes a *very* simple and focused argument using high-information variables. Let me explain. Read More…
For those keeping track, I give you yet one more example of my desire to replace reading text with seeing icons (check the blog’s graphics tag for other examples). Turns out I can trace my fascination with visualizing info to my teen years. My first recollection of being amazed at the visual organization of lots of information was my high school periodic table of the elements. It was a fancy two-side sheet brimming with colorful detail – I can’t find that particular version online, but here’s an example:
In addition to the way in which the elements were organized into periods and groups, other info (in a smaller font, as above) was included as well. One side of the sheet included the atomic mass and many other numeric values for each element, while the other side illustrated some kind of 3D shape/structure associated with each element (haven’t taken chemistry since high school, so I don’t know what it would have been exactly, but I think it looked kinda like the shape of a crystal, or a molecule). I remember being excited, literally excited, at how much information was grouped into that little space – a very efficient micro-macro reference device.
Eventually (as a sophomore in college) I became a historian. What a let down in the visualization department (with the exception of those wacky Annalistes). So let’s talk one of the most frequent ways historians summarize historical info: the timeline. I hope I’m not the only one who hates simple two-column timelines that waste space by spelling out Battle of Mohács, Siege of Vienna… Not only does it take up a lot of extra space, but you can’t include any other info in the list with text alone (who won the battle? which sides were involved? what did it lead to?…). You can of course use a limited number of font style options to add one or two more dimensions to the info in the timeline, but why not use icons, which can manipulate Bertin’s visual variables of shape, orientation, color/hue, value, size, position and texture? Lots of permutations there. This ability to include much detail in a small amount of space is also the motivation behind Tufte’s sparklines, which would be awesome to include in campaign narratives: march rates, casualty rates, army size fluctuations… Yet such things almost never appear in military history, despite our heavy reliance on maps. It’s particularly odd how historians expect these kinds of symbols on maps, but don’t think about using them elsewhere. (Not that I’ve got anything against maps, mind you.)
Thus my mantra: Condense the info and add more, dammit! And, lest you fear, notice that a lot of this info doesn’t even require BIG DATA and massive quantitative datasets – it’s simple nominal and ordinal data.
But back to my timechart-of-the-day (wait for it), which illustrates the major military events of the Ottoman Empire in the first half of the 16C. It probably has a lot of mistakes, but it’s not my fault you see. When you try to create such campaign overviews for most early modern wars, you quickly discover that historians love to mention big events like major battles and sieges, but rarely do they mention the operational context, and they happily skip over entire years when there may (or may not) have been fighting. Half of the time we can’t even be bothered to specify the year: “Several years later the Turks returned…”. Yet even if every combat and campaign doesn’t merit narration, they could at least be included in an appendix or table/chart. (Note to publishers: I would gladly pay an extra $20 for a book that included such basic information, data absolutely critical to judging the quality of an author’s argument. Or I’d pay $20 for digital versions of the kinds of timecharts I post here. Am I the only one?)
So this chart (wait for it) provides yet another example of the challenge non-experts and experts alike face trying to assemble a systematic overview of a campaign based off of spotty narratives that leave out all sorts of details and important information. Reminds me of grad school, where I quickly realized that we could argue about Roberts’ or Parker’s Military Revolution (how much change? when did it change? how did it change?), but since we didn’t know the details of all the various battles/sieges and campaigns, it was a pretty speculative discussion, and not incredibly worthwhile. Now if we had systematic empirical evidence to back our hunches up rather than a decontextualized quotation or two…
In other words, I haven’t found the perfect source that provides Ottoman campaign narratives by year – most of what I have are thematic works by the likes of Rhoads Murphey, Gabor Agoston (apologies for the lack of accents), and Virginia Aksan.
So check out the Symbol page for my icons if you haven’t memorized them by now. And finally, without further ado, I give you the Ottoman wars chart.
So was it worth the wait? I think it works as a good reference sheet, but I won’t say too much about its content, other than to suggest that it does a decent job of illustrating how the Ottomans had to juggle expansion (and defense) along several different frontiers, from North Africa and the Mediterranean, to southeastern Europe and Hungary, to Persia to the Indian Ocean, Persian Gulf and Red Sea. Palace coups, revolts, plagues and fires, raids, battles and sieges – they sure were busy!
Feedback (corrections especially) appreciated. And please make your own timecharts and share them; feel free to use my iconography if it helps (I think it does, needless to say). As usual, feel free to use this chart (no publication), as long as you properly attribute it to this blog.
You gotta admit, this is a pretty cool visualization.
And what secondary source is going to publish something like that? Has published something like that? (Other than a few out-of-print wall-size timelines, mostly from the 18C and 19C.)
It could use a few tweaks – maybe tie it to small multiple maps, or perhaps scale the columns to land area somehow (although the names still need to be legible). But the title, coats of arms and links are nice.
And let’s not forget that Wikipedia allows you to quickly see historical figures’ basic life statistics, all the main titles and offices they held, as well as trace their predecessors and successors with a click of the link. Plus, you can check out the foreign language versions to find more details on certain (foreign) subjects.
You can still find the occasional factual error, and the interpretation in the text will only be as good as the sources the editors cite – but notice how they do use recent, even academic, sources.
Overall, props to Wikipedia I say. Not that I allow my students to cite it in their papers of course.
I figured I might as well add a bit more info to my posts on new books/articles/chapters – mostly because I can, rather than because it necessarily provides a fundamental insight into the work. Starting with our old friend the word cloud:
The above cloud was made in www.wordle.net, and allows for some customization. [For the record, www.tagcrowd.com had a bit more customizability, e.g. displaying the word count beside each word, although it has a text length limit]
Unfortunately, wordle doesn’t do what Google Books does, i.e. provide a list of words that are unique (or at least uncommon) in this source compared to other sources – this would require wordle to have a huge database of documents. And of course Google Books doesn’t include journal content.
If you want to play around with the text a bit more, the free Voyant (formerly the awkwardly-named Voyeur) website provides all sorts of bells and whistles, although it can be a bit buggy at times, especially when you’re trying to use it in the classroom 😦
An example of its features:
As you can see, Voyant gives you a whole bunch of additional info – word clouds (with extremely limited customization), word counts, the original text with chosen words highlighted (“identity” here), the occurrence of each word throughout the work (Word Trends), a KWIC (keywords in context) view of the chosen word, etc. You can also load multiple texts and do comparisons across documents.
One of my other recent software acquisitions (spoiler alert!), Devonthink Pro Office*, provides a few of the same functions, but it allows you to compare any document against all your other documents (a theoretical limit of 300 million words and 200,000 documents per database, I’m told – but you can have multiple databases). I’ll show just a couple screenshots of my far-from-complete system in DTPO. Currently it consists almost exclusively of secondary sources, totaling 14 million words (265,000 unique words), and 7,680 documents.
First, a word count of the article (which only counts words of three characters or more, hence some of the stats are different from Voyant above):
Then there’s a list of words that this article uses more frequently than other documents in my DTPO database use (mostly on the War of the Spanish Succession thus far) – if that makes any sense:
I’m still playing around with how to organize things in DTPO. I’ve been focusing on secondary sources thus far because: 1) it’s helped me the most with my immediate projects, and 2) it’s obvious how I can use DTPO for secondary sources, whereas I still need to give some thought to whether DTPO will be able to replace my Access database or not.
There are lots of other applications for analyzing text sources, but I’ll end with these three for right now. Of course this is just the tip of the digital humanities iceberg.
* Yes, I finally bit the bullet and joined the Apple fan boys, mostly for two pieces of software: Scrivener and Devonthink Pro Office. I absolutely love my MacBook Air BTW (it’s the iPad with keyboard I’d been wanting). I am, however, currently typing on my old PC desktop – will keep both. I’ll post more about my use of them in the future.
I’m playing around to see how easy it would be to convert my Access bibliographic database for published sources into Zotero. It looks like it will be relatively easy, despite the fact that I have a lot of records in Access – a formal count gives 35,448 individual titles, although there are probably a few dozen duplicates. While Zotero has certain advantages to my existing setup, it also lacks a few features as well (e.g. adding lots of custom fields). So I wondered whether losing those features would matter. My first thought: “Well, at least I have most of the sources on EMEMH already in, so it’s not like I’ll have to constantly add new sources.” Or will I? As I quickly realized, one of the advantages with Access is that I can run a real quick query to do a count of how many records I have from each year (publication date). This includes every type of work: journal articles, book chapters, books, edited books, even a few dissertations. Then I can graph it and get the following:
Turns out that there have been quite a few works from the 20th century. Perhaps not surprisingly, their number has increased from the 1980s on. Disconcertingly, their number has dropped pretty significantly in the past few years, although there might be methodological explanations for this, and we can always blame the economy and the decline of academic publishing. For a closeup view of the ‘collapse’:
Other Comments and Caveats:
- This includes the entirety of my bib database, which includes some non-military works, but very few works on 19th and 20th century history (as he stifles a yawn). Since I am obsessively focused on EMEMH and have entered almost all of these by hand, I don’t have citations for a whole range of ‘classic’ (or otherwise) works on early modern European history, unless they directly relate to military matters.
- My bib database includes very few works on Napoleonic warfare as well, the inclusion of which would obviously inflate the 19C & 20C count.
- For edited collections, I included a record for each book, and then separate records for any chapters that were of interest to me. So depending on whether your level of analysis is the book or article/chapter, the count will vary slightly.
- There are an extremely large number of pre-1800 works (22,002) because I mass imported a whole load of titles from a library with individual catalog records from EEBO from the second half of the 17C, which include many books on non-military subjects.
- My bib database includes, IMHO, a pretty full accounting of all the English and French language works on EMEMH (certainly for subjects covering the period from c. 1670 to 1715, as well as late-20C scholarship on all early modern periods) – I’m a bit of a citation collector you might say. Not quite as much for works written in the other European languages, although I do have a fair number of German and Dutch works saved in my Google Books library as well downloaded files saved in various folders (from the KB’s Knuttel Pamflets for example). One of the reasons to switch to Zotero, to speed up entry (even as it slows down entry and recall in other ways).
Anybody know if Zotero can do this kind of querying? I know they have a timeline feature, but I haven’t been impressed with what I’ve seen thus far.
But more importantly, what’s the shape of your bibliography?
I made this graphic back in grad school (in Excel), but since it never made its way into the dissertation or book, and since I’m busy with other things, I might as well post it here as filler. I did clean it up a bit in Adobe Illustrator, just for the record.
This graph shows the length of each siege (by the four stages, as described in my book), in the context of the campaign. The start of each campaign is at the top of the Y-axis, and you read down and across (to the right) until you get to winter quarters and the dividing line between each year. The campaign season is indicated by the white area in between the two gray areas.
Noteworthy items illustrated by this chart:
- Rheinberg 1702 was converted from a short siege into a longer blockade that lasted into winter quarters.
- The extraordinary length of the final stage of Bouchain 1711 (repairing the fortifications) led to some harsh criticism of the engineers.
- Comparing the siege lengths across the years, the difference between Vauban’s pré carré and the Spanish bicoques is quite striking. The 1702 exception isn’t really an exception: Kaisersweert, near Dusseldorf, was on the Rhine and largely conducted by the Prussians, rather than the more-competent Dutch. Amazing what can happen when you don’t bother to fully invest the fortress and allow daily reinforcements and evacuation.
- From the seasons, you can see how late the 1708 campaign lasted (it actually extended into 1709, but I forgot to include that).
- As well as how late the 1709 campaign season started, thanks to the Grand hiver.
- The width of each campaign year has no meaning, other than reflecting the number of sieges conducted in that year.
- I left out 1704 because the main Allied effort was in Germany rather than Flanders, and I focused on Flanders for my book.
- 1707 had no sieges of note.
I could add other bells and whistles, e.g. icons for other events, a few labels, include the 1704 and 1707 campaign years, put the French sieges in as well. Maybe when I have more time.
So what do you do if you teach a variety of early modern European courses over and over (in this case, Reformation Europe, European Warfare 1337-1815, Religion War and Peace in Early Modern Europe), need to quickly get up to speed on the narrative every time you teach it, and fancy yourself a visualizer of historical information? Something like this:
A bit of overkill, perhaps, but I’ve always liked my data dense. I’ve shared other examples of my timecharts before, and this is a more recent version of my overview of the Italian Wars (Wars of Italy if you prefer) in all their nauseating complexity. A topic, it so happens, that I’m covering in class today.
To slightly repeat myself from my earlier posts: this cheatsheet combines information on the names of the wars, their chronology, the combatants involved in any given year, the alliances, the rulers, and the main military movements and combats (battles, sieges) in each war year. I give a copy to students for reference purposes, and display it on the screen as we discuss the narrative of the war. You can also just use the colorful timeline (on left side) as a strip in the margin of a Powerpoint slide if you want to display other material on the slide (you can also trim the columns down to just the main belligerents).
Next up: figure out a way to simplify all this narrative detail down, without dumbing it down. Ideally I’ll add a few maps as well, or at least the same map of Italy with the various alliances, occupations, and major combats as they change over the course of the wars. Now will somebody write a good narrative of the wars in English please? Or even French.
Let me know of any factual corrections, embarrassing omissions, or design tweaks. Be sure to check out the Symbol page at the top of the blog header if you’re not familiar with my symbolism. (On that note, I got Bertin’s Semiology of Graphics for Xmas. Whoopee!!)
This was an earlier version, and I still like the maps (though I need to make my own):
Feel free to use the top graphic in your own courses, with appropriate attribution of course. And let me know! But no publications please (see the Citing the Blog page for general comments).
Perhaps you’re like me. You tend to think about things visually and perhaps after a cartography course and a Tufte book or two you appreciate that visualizations can be far more data dense than an equivalent area of prose. Preferring to think visually is indeed great, except when, like me, you have practically no artistic skills. So you don’t really use it very much because you can’t draw a smiley face, much less a semi-respectable outline of Europe.
Discussion continues on the state of EMEMH publishing – the good, the bad and the ugly. But until we get some more input from others, I’ll shift gears just a little. But please do continue the conversation.
A long time ago, probably when I first starting teaching my own courses back in grad school, I became frustrated with how few EMEMH images I had access to (I should probably start numbering these frustrations for reference, and for therapeutic purposes). This was back in the mid-to-late 1990s, when computer graphics had only just gone mainstream and the Internet was just taking off. It was a heady time, with enthusiastic grad students and military history enthusiasts scanning crappy black-and-white versions of diagrams and maps onto the computer to insert into a syllabus, to use in this new thing called PowerPoint, or to post up on the World Wide Web. A ground-breaking example from one of my most imaginative professor’s courses:
This was in the GIF age, when Compuserve’s image format ruled the world. Much like the dinosaurs, most of these scanned pictures would be consigned to the scrapheap of history within a few years, worthless either because of their poor quality or perhaps because they were saved in a proprietary format that was no longer easily accessible. But that’s beside the point – early adopters always reinvent the wheel from scratch and more often than not proceed to scratch it altogether.
Now that the digital age has truly arrived, we should be able to do better. As I sit here scanning in various battle maps from umpteen different books, I wonder how I will keep track of such things. My normal work flow is relatively set by this stage in my career. For serious textual research, I created a customized MS Access database for primary source note-taking and über-precise keywording, and a related bibliographic database for published (i.e. largely secondary) sources. Thousands of PDF documents are stored in several different folders on my hard drive (some already hyper-linked within the Access database), including hundreds of journal articles and book chapters that have OCR behind the PDF, and are thus searchable using Acrobat’s global search. But my Access database, as powerful as it is, has always been a bit clunky and limited to the desktop. So ever since I acquired an iPad I’ve been using Evernote as an idea diary and multipurpose notebook (automatically syncs between desktop, laptop, iPad…). I’ve even started typing in small quotes and drafts into it, as a temporary holding spot before they get copied over to the formal MS Word document (and possibly back into my Access database). Evernote continues a long history or ‘tablets’ and commonplace books used for hundreds of years: one 17C historian compared this kind of notebook to a fortress, which stored a garrison full of ideas and evidence, any of which could be mustered into the field at a moment’s notice. Evernote admittedly overlaps a bit with my Access db, but it is much more portable across platforms, so it makes it easier when I’m away from the desktop. I’ve also got a thousand books and dozens of volumes of archival photocopies sprawled everywhere, but at least almost all the book chapters/journal articles are now PDF. I’m slowly trying to get as many of these scanned (and searchable) as possible. Google Books is also useful here.
But thus far I’ve treated images like my other teaching resources, with a much more haphazard workflow. I usually end up reading books and articles, and then scanning the occasional graphic and putting it into a folder organized by topics and wars that reflect my course structure. I then use the free graphic management program Picasa (Google’s baby) to keep track of where all the images are, and do a few basic searches in Picasa when putting together my PowerPoint presentations for class/presentations. But now that I’m starting to use some of these images for my research, I feel like I need to be a little bit more organized with my images. So I’ve made a few recent changes. I’ve started tagging the images in Picasa as well, since my file name may say “Landen 1693”, but I won’t always remember to also search under Neerwinden to find all the images related to this Battle-With-Two-Names, nor am I consistent whether I put Landen’s battle plan in the Battle folder or in the 9YW folder – I could make separate file copies for each folder, but I’m quickly running out of space on my hard drive, and any modifications (crop, change contrast…) would have to be done to each image separately. Things are getting even more confused with my recent foray hunting down battle plans from the 9YW & WSS. In order to tell them apart in a file list, I have started naming image files after the exact title of the image, yet this means irregular spelling of names (e.g. Neer-Winde) and sometimes no mention of the keyword at all. It doesn’t help that I’m really lazy when it comes to recording the exact source of the various images I’ve gleaned, esp. online. Fair-use discourages strict citation practices.
So, how do I easily keep track of all these?
At this stage Picasa is the default, but it lacks the various metadata features available in other programs, particularly my Access database. Another solution would be to integrate it into my Access database, essentially treating each image like a book chapter or webpage. Perhaps another route might be to use Zotero, since many of the images (or their bibliographic records) are already online. Or possibly paste copies of the graphics into Evernote.
In theory, there might even be some way to quickly identify the location of all sorts of illustrations, using the Google Books API to extract the items in the various List of Illustrations, Maps, Figures in published works. But the technical skills inherent in that previous sentence have already put me in over my head, so I’m not the man for the job. Not to mention, I don’t really want to add yet another application into my workflow: Access, Evernote, Adobe Acrobat, Word, Picasa, Zotero…
Given time constraints, most likely what this means is that I will continue to use multiple programs for their niche features. Ideally there’d be an all-in-one program that is customizable (i.e. can create your own queries and see the backend data), easily handles text and image, scalable (we’re talking gigs of data), exportable to other programs, and is available and syncable across multiple platforms. Talk about your non-existent digital chimera.
Recommendations for combining textual and graphical items into an integrated workflow?