When we started we had four large boxes of old and fairly brittle index cards. The first thing we needed to do was transfer these from the physical world to the digital, in other words we had to get them scanned. We could have done this by hand on the old manual scanners the archive had but that would have taken quite some time as there were around 4,000 of these cards with detail on the back and the front. So, we sought help from other quarters: did any of the museums have facilities we could use? Not any that we contacted. Any commercial organisations fancy offering any help? Not the ones we tried. Thankfully an internal source came to our rescue.
We were given a contact at HSBC Coventry Scanning Centre as someone who may be able to help. They operate a number of industrial scanners that capture any paper forms and letters sent in by customers. She believed we could use these to capture the cards.
Her initial assessment looked bleak however in that the scanners were tied to a bespoke bank system and there did not seem to be a way for us to get the images out. Damn.
However a week later she got back in touch saying she had cracked it! Yes. She had found a queue that would allow images to be extracted for use elsewhere. Then all we had to do was prove that the big automated scanners they use would not damage the War Cards. A testing facility for the scanner exists in Sheffield so we ran some tests there to prove the cards would not be torn to shreds by the scanners and thankfully all went well and we could proceed.
A few weeks later Rachael and I headed to Coventry, Rachael driving from the south with four big boxes full of cards. We were helped by a number of staff who had the task of capturing the 4,000 cards over the next few days. Off we went scanning the cards in batches of 30 at a time. We worked our way fairly slowly through the first box and then found we had been capturing the images to the wrong queue i.e. one we could not extract images from which meant we had lost a mornings work and had to start again (annoying but it could have been much worse).
Two days later the images were captured. A few days after that we had them on our internal network in a series of named folders each containing a batch of 60 images representing the front or back of a War Card. We were off and running. We can’t thank the Coventry team enough for the help as without them this would never have happened.
Finding a Platform
The next phase of our plan was to build a crowdsourced transcription platform, which is basically a way to host all the images and get lots of other people to type in the data on the cards. I had posted publicly on both our internal blog (this one) and my external blog about the War Cards and asked for ideas and suggestions for suitable software. Thankfully a few people got in touch with ideas for help. A lot of people were very helpful including but not limited to;
- Ben Brumfield from the US is an expert in manuscript transcription and even had his own software which we experimented with for a while but did not quite give us what we wanted. He gave us lots of help and pointers which were fantastic
- Kim Plowright gave us lots of useful pointers and names and at the time she was herself involved in Channel Four’s subsequently award winning D-Day events.
- Chris Thorpe was also involved early on and again was very helpful in thinking about transcription challenges.
- Nick Jewell provided a fantastic and detailed comment to the initial blog post internally suggesting a great set of routes to take including looking at commercial platforms like Zooniverse5. We did and it was a bit big for our needs but I did meet some nice astro physicists).
- Simon White suggested this lovely looking open hardware book scanner6 which if this project is successful enough someone generous might fund for us to use in future projects. Hint, hint.
Finally we were also contacted by Luke Smith from the Imperial War Museum who was running a very interesting project called Lives of the First World War which sounded like a perfect home for the War Cards and he was very keen to have our unique data set. We just needed a platform to convert what we had into something the Imperial War Museum’s system could consume.
My colleague Paul Dougan came in to his own at this point diligently researching archival standards such as Dublin Core, doing this right and that was a key requirement for our platform and he was instrumental in testing potential software solutions of which we tried several. Ben Brumfield’s platform ‘From The Page’7 was a bit too free format. We also trialed the Omeka8 platform, which uses Media Wiki the same platform that Wikipedia is built on, and has been used by several other historical transcription projects and looked for a while like our likely solution but we could not bend the UI to our will.
After a few false starts, red herrings and failed tests Paul made the decision to build something from scratch. He had wanted to use Drupal9 a popular open source CMS for quite some time and this gave him a good excuse. Drupal has loads of support, loads of modules and loads of flexibility, it looked like something far more able to meet our niche needs. Quite quickly Paul had a working prototype built and we had the beginnings of a useable and configurable system. The system was constructed without the need for any “coding” but through the configuration of preexisting building blocks. The only proper code being a small script to assist with the data import/export and validation processes.
But where would it live?
My initial thoughts on this were that the transcription system would be hosted on the web. Oh how naive I was. We thought we could just get an Amazon Web Service instance up and running, host the system and let people loose on it. Sadly not.
While the War Cards contained details of staff sadly no longer with us the data is still treated with the upmost importance which meant we had to abide by existing legal and information security standards. Anything hosting bank data needs to be rigorously penetration tested by an external security firm. This would have incurred a charge of tens of thousands of pounds which would have crippled the project.
We would also have had to use approved 3rd party hosting providers – the systems and processes around it were not really geared up for these kinds of experiments. There were also legal and empathetic concerns which we could not check before transcription e.g. would the data contain anything that could be embarrassing to living family members? There were challenges around registering users, catering for double entry to ensure accuracy, as well as implementing solutions to prevent mischievous transcription efforts.
We decided it would be easier for all concerned to do this inside the organisation on our own development servers. Once we had made that decision it was relatively simple for Paul to get Drupal up and running in our environment. We were well and truly on our way (it sounds relatively easy but these simple sentences don’t convey that it took us several months to get to that point)
The system takes shape
By the end of January the system was ready to start testing with real transcribers. Paul had worked tirelessly on making sure the data was captured and stored correctly within the system and that the transcription data could be extracted accurately after the fact to handover to the Imperial War Museum (IWM).
The system needed to not only deal with the transcription but also had to allow for multiple user profiles and access to the cards without locking, feedback mechanisms, administrative features that would allow the archivists to approve the transcriptions, as well as statistics on where we were in the overall transcription. By the end of the project it was a worthy piece of lone engineering.
The main transcription screen of our system
There was a conscious decision taken to build a minimal viable product and to start using it before it was entirely complete and to iterate continuously in response to user feedback. In fact we received over 700 pieces of feedback from users relating to all aspects of the system although mainly these referred to the legibility of the handwriting on the cards.
We loaded up our first batch for testing and the three of us transcribed some cards. Rachael is a trained archivist so reading the old script was a piece of cake for someone with her skills. Me, not so much, for someone with such awful handwriting you would have thought I would be good at reading what looked like scrawl…but no. Thankfully there are quite a few typed cards.
Test yourself on the War Card of Monty Parish and see how much of it you can read.
Would you have made the grade? What did you think his first name was?
Read on to hear more of the story contained within this card.
Getting people into the trenches
After our tests we were ready to open the doors wider and we had been teasing the launch of the system via our teams internal newsletter and blog, a War Cards Community site and also on our internal collaboration back channel, uBlog (An internal Twitter clone based on Statusnet). As such we had a group of around 20 volunteers who could help us gently kick the tires of the system and ease our way into the transcription effort for real. The first few batches of images were transcribed in just a few hours. People were talking about the stories on uBlog. Problems and enhancement requests for the system were being raised and it was clear fairly early on that this was going to work.
We added more images, and we opened the doors publicly; we invited more people through our own networks and we started to build transcription momentum. We received a further boost when our project got a mention in a couple of widely distributed and read newsletters at HSBC. The week after those letters went out we had over 100 people helping with the transcription effort. With around 8,000 images in need of transcription we needed as much help as possible.
We certainly got it. The word began to spread organically and by the end of the campaign we had over 274 transcribers registered from all over the globe. The pace of transcription picked up dramatically throughout the transcription period as we added new people to the cause. We started to see 5-10% of the cards transcribed per day at our peak output. Every little bit helped and we are very thankful for everyone who transcribed even a single card.
Of the 274 transcribers we had some real superstars, people who went above and beyond for the cause. I would like to make a special mention to our top ten transcribers for the War Cards effort. They contributed 1547 transcriptions which is just under 40% of the whole set. Absolute heroes.
The first batch had gone live on the system on 11th of February and the final transcription batch was completed by the 2nd of May. We loaded the batches quite slowly at first while Paul tested the system for leaks and cracks. When batches of cards would run out we had people clamouring for more.
We had some more difficult batches as the final hurdle. Some of those had stories more detailed and complex than a single card could hold. The longest of which ran to six cards back and front. These needed a more bespoke solution to which Paul again dealt with.
Bring on the approvers
Once the transcription was complete to a high enough level (once we reached around 90% transcription) then the work had to be approved. The rigorous standards of the archive team needed to be met. Every transcription needed to be approved. Any transcription errors and flagged issues also need to be fixed. The archives team was just eight people so by no means a small task. The first card was approved at the end of March and the final one was approved by Rachael on the 11th of June.
The archives team soon found that approvals wouldn’t be as simple as first thought. Whilst the transcriptions were great, there were lots of small simple amendments to make, such as consistency, the spelling of overseas places visited by the men, or battles they fought in. Google was certainly a friend during this process. Members of the archives team lined up ready for duty, as Rachael gave batches to archivists in both the UK and the US to get through and, much like during the transcription phase, people were battering the door down for more cards to approve. It was clear the project was really capturing the imagination of the team, even at this potentially monotonous at times approvals phase.
Throughout the approvals process, more tweaks to the system needed to be made. There was a great deal of feedback which had come in from the transcribers, and Paul worked out a way of clearing this and approving the cards at the same time so the archivists could kill two birds with one stone. The team also needed to keep a beady eye out for any cards that posed a potential risk, from a legal point of view. An agreement had been made with the Legal team that the cards posed very little risk to HSBC, but more that we should be checking for any upset the information could cause for living relatives. So the archivists watched out for details of hereditary diseases or indiscretions…of which there were a small handful but nothing so serious that our legal team would not clear them for release.
And with that the approvals were complete. We had a fully transcribed data set ready to be gifted to the Imperial War Museum.
Handing over the data to our allies
Once the data set had been approved on the 11th of June Paul then had a lot of work to prepare the data (he had to make many hundreds of edits and fixes), adding the cards with multiple backs (this was a real pain), manually reconciling records, merging cards and dealing with anomalies. Painstaking preparatory work. By the 2nd of July it was complete and ready to be handed over for loading.
The data set was loaded on to the Imperial War Museum platform over the weekend prior to the 100th anniversary of the start of Britain’s involvement in the First World War, the 4th of August. Unfortunately there were a few issues. Only the front images were loaded initially and there were some issues around free access to the data set. Thankfully these issue has now been resolved.
These problems meant our launch was a bit of a damp squib on day one i.e. it did not go quite as planned, but the main thing is that the data is now available for all to read and use.
You can find the Midland Bank War Cards at the following address.
https://search.livesofthefirstworldwar.org/search/world-records/midland-bank-roll-of-honour-1914-1920
Screenshots from the Lives of the First World War platform
The full data set to browse and search against.
The transcribed record of Frank Sykes
The front image from Frank Sykes’ War Card.
Please do register for the service, peruse the data set and read some of the stories for yourselves. Weave the profiles into the stories that are evolving on that platform. Help connect the dots so this data forms an important part of Britain’s history. If you need further motivation to participate, and you should not, then the next section will hopefully help with that.
The feeling of transcription
The above details how this project and system came into being but the real interest lies in what was contained on those cards. There will be more written about those within over the coming months but I wanted to add a few words on the effect of those cards on myself and others.
From my own point of view the first feeling you have when reading these cards, and I read a hell of a lot of them during the scanning in Coventry, is humility. You cannot imagine what these young bank clerks and managers would have been feeling after being plucked from their day jobs and sent for training and then to the war. Transcribing these cards on a Monday morning gave you a wonderful perspective for the week ahead and how small your challenges really were in the grand scheme of things.
Of the cards I did transcribe the bleakness of the card for R. Kemp10 really stood out. A typed out card, the back of which told a very short and sad story.
“1918. Spring. Crossed to France March 22nd. Wounded in the back and taken prisoner by the Germans. 24.8.18. Heard of his death.”
It is this minimum of detail which makes these cards so valuable yet also deserving of so much more. I have been unable to find out any more details for R. Kemp.
Another observation was the amount of people that did return to work. Obviously these were young men so they would have needed to return to earn a living but I cannot imagine what it must be like returning to a world of ledgers and papers after the horrors of trench based warfare. Of the 4,000 men that went to fight over 700 of them never did return.
One of the more joyous parts of this project was how other people reacted to it. We had a feeling we were doing a good thing and it was nice to have that confirmed. As part of the volunteering registration Paul made the great decision to add a field to allow people to say why they were getting involved. We did initially think this field would be largely ignored but we were proved very wrong. We got a multitude of great reasons and stories for people getting involved and you could feel a real personal engagement with the project.
“My Grandfather and Great-Uncle fought in the First World War and although my Grandfather eventually died of his war wounds both of them made it back to their families. This is in their memory and for their regiment and all who fought from Yorkshire.”
“I had two relations who fought and lost their lives in the war and recently visited their graves in France. It was through someone making this information available on the internet that I was able to find their final resting places.”
“My great grandparents met during WW1. He was a soldier and she was a nurse who treated him and they were subsequently married for over 70 years. Always been fascinated with the period ever since they told me the story when I was a little girl.”
From an archivist’s point of view
Rachael had this to say about one of her most memorable stories from the War Cards;
‘One card which struck me as I worked on the project tells the story of Frank Sykes, who was in the Field Ambulance section. The remarks on his card show that he’d written back to his colleagues at the Midland Bank branch:
On the 28/7/15 he wrote:
“at last we are off, or shall be at 11 o’clock tonight. My address henceforth will be Mediterranean Expeditionary Force”.
In his last letter dated 6 Oct 1915 he said that he had been in hospital: “nothing serious, generally run down”. I believe the cause was hard ambulance work, under very trying & bad conditions. From the descriptions in his letters it must have been a terrible time for him & others. But he was always cheerful.
“Mr Sykes died on the 24th Nov 1915, whilst on active service in Gallipoli”.
That was the official notice. By his death, the Bank lost the services of a splendid & faithful servant’
The reverse of Frank Sykes’ War Card
We have been able to find more details of Mr Sykes and we have added them to his life story which you can see here
Most of the people involved in transcription of the cards can remember a certain card or elements of a story, the limited details they do contain are still enough to ingrain the person or the story strongly within in your mind.
Wider coverage of the stories
One story which has received wider coverage is that of Midland Bank employee Montague Parish11 and his brother Stanley. The story is told by the son of Montague, John, who was also a Midland Bank employee. He is 88 years old at the time of writing and he was interviewed about this incredible story.
“Monty was moving forward in a bayonet charge when he was hit by a sniper’s bullet – and later described it as being like ‘walking into a wall’. The soldier laid down in no man’s land – before, unbeknown to him, his brother Stanley was asked to take the last stretcher and bring back a ‘wounded officer’. This turned out to be Monty.”
https://www.youtube.com/watch?v=cnLpvUJfOCo
You can see the full interview with John on Youtube. The story also received wider coverage in the Daily Mail.
These are only a handful of examples of the stories contained within. There are stories that end happily as well as those that do not, there are those that are beset with the kind of luck you would not wish upon your worst enemies and there are those that you will believe have more lives than a very lucky cat. Contained within are a good deal of people who were decorated for their efforts. A few Military Crosses and even a Croix de Guerre. Please take a look for yourself and see what you can discover.
What next for the War Cards?
We have this incredible and unique data set. We are hoping the Lives of the First World War will unearth some fantastic stories and add extra details to the lives of the men from Midland who went to fight. Please do contribute to the project if you can.
We will be forming a community on the Imperial War Museum’s site to help build up the stories of the Midland Bank staff contained within the War Cards.
We are also not done with the First World War. As part of our investigations we also looked into the memorial to the war dead at the HSBC HQ in London. We are looking to cross reference the data sets against those remembered in stone. One man mentioned on the HSBC memorial is remembered at a local memorial in Totley, Sheffield less than a mile from where I live today.
We also have plans for the data set for our own usage. We are keen to hack away at the data and see what stories, patterns and details we can pull out. For example we are keen to visualise the geographic spread of the call up, to show the impact on the then workforce (at the time of the war Midland employed around 13,000 staff, these cards represent 4,000 staff members going to fight).
If you have any ideas for things you would like to use the data set for then please get in touch. We have some data hackathons coming up later in the year and we are always looking for capable volunteers. We are also thinking about hosting the data set publicly elsewhere in its raw form.
We hope that the techniques we have used for the above project will stand us in good stead for future projects, with the 150th anniversary of HSBC approaching there is certainly scope for other similar challenges from the archives. Our archives do hold some wonderful historic treasures and I would love it if more people got to experience them. Currently, they are mainly stored in physical form in specially designed buildings across the globe. For example I know that in the UK Archives they have six full height filing cabinets’ worth of photographs. I would love to liberate those.
If anyone reading this has the budget and desire to fund a project to free some of those records and make them available to a wider audience, especially to purchase some professional scanning equipment, then please do get in touch. The open source hardware book scanner would be a lovely thing for us to own.
Conclusion
From Rachael’s first mention and my first viewing of the War Cards it was obvious we had to do something with them and not leave their stories held within four cardboard boxes stored in a warehouse in London. Thankfully others felt the same. We think we have achieved that.
I have thanked most of those involved in the words above, I must also thank Jen Goodison for reviewing this story of events and also to Emma Fahy in the Media Relations team for her help in getting this project wider coverage. I hope I have not missed anyone out. It truly has been a fantastic project to be involved with. The help we received along the way was exceptional and was part of what made it so special.
The majority of the credit must go to two people. Firstly, Rachael Porter for her dedication to this unique dataset and the desire to do something meaningful with it. In Rachael’s own words
‘This project has been an absolute highlight of my time at HSBC. Having the opportunity not only to make such fascinating records available to more people than ever before, but also to work with some brilliant people, has been a delight!’
Secondly, Paul Dougan who made the system a reality, transcribed a large percentage of the cards and ensured the data got to the IWM in a state that allowed it to be shared with the world. His efforts cannot be underestimated especially remembering that this was effectively a ‘side of desk’ project.
“It’s the only thing I have done at the bank that is likely to be around in 100 years”
Finally Rachael received an email on the day we finished this article from Edwin Green, the original finder of the War Cards after he heard about the project.
“This is simply wonderful work. I suppose that the boxes of cards were a lucky find all those years ago but what counts now is the huge extra value and context which you have added to the records.”
All in all it has been an inspirational project to be involved with. I hope our work has done justice to the men who went to fight in the war and that through this project previously unknown details of their lives will help their memory to live on.
You can find the Midland Bank War Cards at the following address
https://search.livesofthefirstworldwar.org/search/world-records/midland-bank-roll-of-honour-1914-1920
Please go and build stories of their lives so that others will know.
Aden, Rachael & Paul
Links:
1. Original What to do with the War Cards post
http://www.adendavies.com/what-to-do-with-the-war-cards
2. Transcribe Bentham Project
http://blogs.ucl.ac.uk/transcribe-bentham/
3. Old Weather Project
http://www.oldweather.org/
4. Lives of the First World War
https://livesofthefirstworldwar.org/dashboard
5. Zooniverse – The Zooniverse platform is being used by Operation War Diary which is an ambitious project to transcribe all the operational war diaries written in the field during the conflict. https://www.zooniverse.org/ & http://www.operationwardiary.org/
6. DIY Book Scanner
http://www.diybookscanner.org/
7. From The Page
https://github.com/benwbrum/fromthepage/wiki/Preparing-a-Work-for-Transcription
8. Omeka
http://omeka.org/
9. Drupal
https://www.drupal.org/
10. R. Kemp’s War Card Transcription & Images
https://search.livesofthefirstworldwar.org/record?id=gbm%2fmidland%2f1885&highlights=%22%22
11. Osbert Montague Parish’s War Card and Life Story
https://livesofthefirstworldwar.org/lifestory/3383870
