Building a Map for 1916 Log Book
Having dealt with the primary source I would be animating as part of my internship with my last post, this post will look at similar maps that inspired my project, my experience building a map, plus some obstacles I faced and how I overcame them.
First I felt it was important to research similar map projects and steal from them. The first such animated map I encountered was via my internship supervisor Dr Mark Sweetnam in one of his lectures about visualisations. He gave as an example a taxi visualisation posted on the Mashable website. Pulling the GPS coordinates from taxi sites like Uber and Lyft, which actually use those coordinates to measure a person’s fare, Chris Wong was then able to make a very attractive app that followed several New York City taxi drivers over the course of 24 hours. It was also able to notch up the number of customers, their fares, and the time each journey took. Google Maps have recently started doing this with their new API Google Roads. Now with smartphones as people move around they drop GPS coordinates, called “crumbs” as in the story “Hansel & Gretel”, which can be formatted in a map and visualised – useful for car journeys, cycles etc.
The next map that was of enormous use was a fan-made map of Westeros, the fictional setting of Game of Thrones. An article in the Guardian first drew my attention to the project, which impressively mapped a fictional world. How they achieved this was by referring to the officially authorised map however it isn’t totally accurate and some differences exist. Similar looking areas on Google Maps were pulled and then overlaid in a grid to form the map. This overlaying directly applied to my project, however I would be overlaying historical rather than fictional maps.
Also relevant to my map was the progress bar. This progress bar charted the movements of the characters through either the books’ chapters of the TV series’ individual episodes. When forwarded they emulated the taxi marker mentioned above, stopping at either close of chapter/episode or when a character met their end. As with the NYC mapping project the developers posted their workings and methods on a blog.
The Down Survey Map project was also very inspiring. In addition to providing wonderful examples of how to georeference and overlay older maps on top of Google maps, the project’s site also contained useful information about how to use historical GIS (Geographic Information Systems), databases for storing information, and also the user interface. The spirit of pulling help and inspiration from the Down Survey project would continue when my supervisor put me in touch with David Brown who worked on the project. David was very knowledgeable on the subject of online maps, and advised me on how best on how to meet my project milestones, which I’d been quite ignorant about up to that point. I will return to his advice, which concerned introducing time to the maps, in the next post.
Before starting into the map-making I had one or two further research questions to answer. This would be a useful exercise to try and avoid setbacks because I hadn’t queried the dataset sufficiently at the start. Nonetheless the overlooking of some salient information contained in the logbooks would prove to be inevitable.
These research questions included:
• What exact route did the ambulances actually take?
• Where did the ambulances go to exactly?
There was no guaranteed way of telling what route the ambulances took, crumbs of GPS coordinates not existing back in 1916. This fact threatened to scupper the whole project, because as someone pointed out, the Irish history community would be up in arms if the information being visualised turned out to be false. Irish historians were after all one of the target users.
One solution I put forward was to adopt an as-the-crow-flies approach and have the marker representing an ambulance sailing above buildings and streets. While this would have been safer in terms of historical accuracy, it lost much of the appeal and interest that a marker navigating Dublin during Easter 1916 would have, it actually being very compelling watching a marker snake through the city, as seen with the NYC taxi visualisation.. I include here the alternative map.
Such a plan would have made short work of my internship project. The only tenable alternative was to research what would have been the most likely route to take, and attempt to make it as historically rigorous as possible. This required considerable research on the part of Finola Frawley at Dublin City Library, who delved into the history of events during 1916. I assisted in a small way myself with the scholarship. What we were looking out for were the locations of the volunteer strongholds, barricades, and British garrisons, reasoning that these would have influenced the route of the ambulances, rendering those streets impassable or inadvisable to drive down.
I marked this information onto a screenshot of a Google Map of Dublin through PowerPoint, as a useful reference to what route the ambulances may have taken. I include them below, as well as a scan of a hand-drawn map containing most of the military locations.
Next the question of where the ambulances travelled to exactly. This was necessary to answer because while the ambulance logs would list the area and sometimes the phone number of where a call came in from, the precise address was rarely provided. What this would have entailed was that the ambulance marker on the map would turn onto the specific street connected to the call, and then suddenly stop dead, which would have been extremely unrealistic.
Working together, Finola and I tracked down registered phone numbers in Dublin. Pulling them from a website named Lemon Wylie, I placed all 5,963. phone numbers on an Excel sheet, meaning I was able to search the phone numbers for the exact listed address. However some limitations existed: because the website only contained phone numbers from 1913, all the phone numbers stopped before reaching the number 4000, and several of the phone numbers listed in the log book number were well above that. Another consequence of being three or more years old is that while the addresses are accurate, many of the placenames or business had changed ownership. This method was undoubtedly useful for determining the exact address of each call, but was not a reliable way of gaining information about the callers (concerning locating the address of the numbers above 4000 the same site offers access to the 1918 phone directory which can be used to place them).
This method solved the problem of how to pinpoint the address of most of the phone numbers listed in the log book (also it should be noted that many of the phone calls were unattended due to military action). But there are other numbers that appear in the log book eg “68 Fire Alarm” or “box. 68” that do not correspond to phone numbers or addresses.
These are actually emergency fire boxes that had direct lines to the nearest fire station. Whenever an emergency was detected a civilian would use the box, their name would be recorded in the log book, and they would have to remain at the scene until the ambulance arrived. In an interview with the Dublin historian and fire brigade expert Las Fallon that Finola conducted, Las explained the role of the emergency boxes: “[I]f there was a fire or an incendiary you ran up & you pulled the box and you waited there and the fireman arrived and you said, ‘it’s 100 yards down there and up to the left, it’s a house on fire’.”
Finola set about investigating these fire boxes, and unearthed many interesting pieces of information. I include below a transcription of an article Finola found detailing the intended construction of fire boxes in 1902. She found other archived materials relating to later fire boxes, which helped to conclude where it was the fire boxes were located. We were not able to match the box numbers directly with the fire boxes, too few exist anymore to allow this. But because of how fire boxes were conveniently located – on the corner of one street and another – placing the fire boxes turned out to be quite straight-forward. Included below together with Finola’s transcription is map of some of their locations.
Now I was ready to begin making the map, and needed to decide which map engine to use, or would I invest in proprietary software. Google Maps was far and away the largest and most widely used maps engine, and appeared to be used by all of the similar visualisations I’d identified and resolved to steal from. The Game of Thrones map of Westeros combined several of the features I hoped to incorporate, all achievable on Google Maps. More visually slick was the cool NYC taxi visualisation, which was largely made using JavScript and JSON, and hosted on a map site called MapBox.
Meanwhile I was researching other viable mapping options, which were Esri ArcGIS and NeatLine. Esri was a paid proprietary mapping tool but I downloaded a free trial. The controls were more straightforward than Google Maps (very trial-and-error requiring writing source code), however Google Maps technology looked and felt better. However Esri remained an alternative should the Google Maps API suddenly become unworkable.
The below screenshot is of a map of Dublin with a marker placed at Tara Street Fire Station. On clicking the marker an info window pops up containing text I pulled from Wikipedia and a stock image of the station. It is Google protocol to limit the number of open info windows to one at a time, so as not to overwhelm the map.
Next it was necessary to chart the route that the ambulance took. In order to do this I took as a fictional route from Tara Street Fire Station to the General Post Office, unlikely to have happened given the GPO’s occupation by the Irish volunteers.
Once the animated path was achieved it was then necessary to figure out a way of placing the 1916 map of Dublin over the modern-day Google map. This led me to Overlays. It is possible to overlay any image on top of a map, and have other features at a remove to the base map, like markers and paths, appear on top of the overlaid image. The image must be rectangular however, because the north-west and south-east coordinates of the map image must be given. As can be seen from the below example, while this is generally effective it is not 100% accurate. In contrast Google Earth requires the latlng of the four corners of the image, improving accuracy. It could be also that the image being overlaid isn’t entirely straight either.
To finish, many questions remained unanswered. How would I represent time in the map? How would I more accurately overlay a historical image onto a map? How would I link this to a database? And with the code going into Google Maps becoming increasingly difficult, would I have to seek another route?