Thursday, July 9, 2015

‘and the war came’: Visualising the human costs of the American Civil War (1861-1865)

[Don't want to read my waffle & just want to get to the Viz? You'll find it here or you can just scroll to the end of this post]

Some time ago I completed a vizualisation of the human expenditure in the First World War (1914-1918). It was a pretty satisfying experience that allowed me to deepen and broaden my own understanding of the period and the finished product will, I hope, allow others to do the same. I was casting about for another similar undertaking, both as a means of enhance my skills with the Tableau application and act as a training platform before I undertook a larger data vizualisation project based on my IR&DD Catalogue of Irish radiocarbon dates (operational Functional Test available: here). Over the last while, my long dormant interest in the American Civil War has been revived and revitalised by a Damian Shiels’ rather wonderful work on the Irish involvement in the conflict [Blog |Facebook | Book]. As the blurb on his book states, with almost 200,000 participants of Irish origin, it deserves to be recognised as one of the most significant military events in Irish history. Thus, it came to me that this should be my next project.

Using the List of American Civil War battles Wikipedia article as my guide, I collected data on the following: Belligerent, Theatre, Battle, Date (Start & End), modern US State, Civil War Sites Advisory Commission (CWSAC) designation (where applicable), Strength deployed in the field, Casualties & losses, Prisoners, Victor, Additional Notes, Wiki Page address, along with Latitude, and Longitude. The list of Civil War battles is composed of three parts: Battles rated by CWSAC (385 actions), Other non-Indian wars battles and skirmishes not rated by CWSAC (55 actions), and Other battles in the Indian Wars (10 actions). I have included all of these in the present data set.

As with the WWI data, I’ve had to address a number of issues in one form or another to allow the information to be fashioned into a useable format (I would suggest that the interested reader consult the ‘notes on the data’ section of the WWI post, as many of the issues are identical). The most common of these is the variability of the data when it comes to the strengths of the various forces, and the losses they suffered. Where a range is given (e.g. 1000-1500), I’ve consistently gone for the lower number. Where possible, I’ve attempted to separate out the numbers of prisoners captured. However, owing to the variability of the data, I’ve had to lump all casualties (both dead and wounded) together. Although imperfect, where forces are expressed in military units, I’ve taken a division to be 10,000 men, regiment to be 1,000; a company to be 100, a battalion as 300, a corps to be two  divisions (c.20,000) and a brigade as three battalions (roughly 1,000 men). I, of course, realise that this is a deeply flawed exercise as expected unit sizes were frequently quite fluid during this period, coupled with the fact that most never reached full strength during the entirety of the conflict, but it will (at least) suffice to give some broad outlines.

How to use the Vizualisation
Whether you use the embedded version of the viz at the end of this post, or go directly to the Tableau Public server, you'll see the same landing page giving an Overview of the content.

Overview of the Dashboard
This dashboard is divided into four panes and shows Battle Site, Strength, Losses & Casualties, along with the Missing & Captured. In all four, the dots are coloured by the victor in the encounter (including if the outcome was indecisive). The controls on the right of the dashboard provide the first level of data selection. With these, the researcher can choose the Theatre of the conflict, the Victor, and a calipers allows selection of battles based on start date. It is important to remember that all choices made here are automatically applied to all subsequent dashboards. The following four tabs (accessed along the upper edge of the viz) replicate each of the four small maps on the Overview tab. As before, the Battle Site tab uses a single dot for each encounter, while all of the others are resized by the numbers of people involved. 

The Losses & Casualties tab
On each of these tabs the user can discriminate further, refining the Theatre, start date,  and Victor. On all but the Battle Site map, there is a scale to show the relative sizes of the actions in terms of the strengths committed to battle, the dead and wounded, or the missing and captured. However, the State control is added here for added ability to interrogate the data. For example, you may be interested in battles that occurred in North Carolina, but not care so much about what happened in Florida. Essentially, you can select all the data that's relevant and of interest to you. It should also be noted that additional choices made here in selecting or deselecting data will again filter all of the other dashboards that use the same dataset. It shouldn't be an issue, but even if you can't get the Viz back to its original state, there's always a 'reset' button of each page. The other thing to note is that hovering over any of the points will bring up a 'tooltip' box with additional information, and clicking on any dot will open the relevant Wikipedia page.

Suggested further reading from Wikipedia
Once you've made your selections and read the 'tooltip' notes you may care to do some further reading an analysis on your own. The 'Further Reading' tab has you covered! Based on the map data, the battles in your selection are listed alphabetically, along with the Theatre of conflict and the name of the Wikipedia page where I got my data, and a small mark to indicate the start date of the action - this should give the user a 'ready reckoner' to distinguish the broad order of the battles. Clicking on any one of these (text or marks) will open the Wikipedia page in a new tab. The final tab, Order of Battles, doesn't really work as well as I'd hoped and expected. In the Tableau development environment you can create an animation that plots through your selected data on a day-by-day basis showing where each battle occurred. I had expected that this would translate seamlessly through to the version available on the internet (well, everything else has!) ... it has not. As it stands, you can manually scroll through your data subset one day at a time, or jump to any battle start day in the data. It's not ideal and I had contemplated removing the dashboard entirely, but it may still be of interest to some, so it remains. While my hope is that the good folks at Tableau will remedy this situation, I realise that the memory usage to achieve this 'live' over the internet is prohibitive. But I live in hope! On the other hand, the video below will bring you through every action in the US Civil War - from Fort Sumter to Palmito Ranch - and back again ... all in under three minutes!

All the battles of the US Civil War ... in order ... forwards & then back ... in under 3 minutes!

The first part of the title of this piece is taken from Abraham Lincoln’s Second Inaugural Address, given near the end of the war in March 1865. In searching for a title, I was reminded of it by my friend, Ed Feeny. Even after all these years, it still holds a stark, mournful tone, recognising the losses on both sides of the conflict.

In writing this piece, I've become aware that there is a difference in how the the word Theatre is spelled in the UK and in the US (Theater). While I'm mildly sorry if it causes confusion or annoyance,  I'm nearly positive that this will not prove fatal.