Military History from the Street: New Methods for Researching First World War Service in the British Military

I. Introduction

A central element of British popular memory of the First World War is the idea of communities and interest groups enlisting en masse and going off to war together in infantry units with some specific characteristic. That idea can be found most clearly in the stories of the ‘Pals’ battalions which drew so heavily on close-knit areas that the devastation wreaked by losses on the Somme was immense. Equally, it can be found in the volunteers of the 17th Middlesex Regiment, the ‘Footballers’ Battalion’, ¹ or the Cambridge undergraduates embodied by the fictional Lieutenant The Honourable George Colthurst St Barleigh in Blackadder Goes Forth. ² Remembering August 1914, he described: ‘Myself and the fellows leap-frogging down to the Cambridge recruiting office, then playing tiddly-winks in the queue. We’d hammered the hell out of Oxford’s tiddly-winkers only the week beforehand and here we were off to hammer the Boche.’ ³ Questioning from Captain Blackadder rapidly reveals that the lieutenant is, by 1917, ‘the only one of the Trinity Tiddlers still alive’. This narrative of slaughter, in which only the lucky few came home, has been unpicked by countless revisionist writers, most notably in the context of popular culture and memory by Corrigan, Hanna, Sheffield, and Todman. ⁴ In this, academic opinion is even more than usually distant from popular memory.

However, historians and the public seem to be more united on the issue of local communities’ service in the military. At the core of writing about military service in Britain and Ireland during the First World War is an implicit assumption that the units in which men served generally had some link to their ‘home’ area, or at least the area in which they enlisted. This has led to several different historiographical genres which focus on the links between units (often infantry battalions or the infantry in specific divisions) and geographic areas.

In the first place there is a well-established tradition of writing about Pals battalions. ⁵ Such texts include significant material on the local character of each battalion, local efforts at recruitment, and the ongoing relationship between local areas and ‘their’ battalions throughout the war. Second, where divisions had a specific geographic link, their histories often include reference to their recruitment areas and the circumstances in which they were raised. ⁶ Third, highly localized publications, which seldom reach beyond local history societies and a few bookshops and museums, tell stories in a variety of different ways. They might be rolls of honour, which are primarily listings of those who died, ⁷ or narrative accounts (but again focused on those who died), sometimes focused on a selection of units but more often on one or two battalions with local links. ⁸ Some publications combine both approaches. ⁹ Fourth, there are now academic analyses of the relationship between local areas and military service. Such publications which deal with the UK concentrate on battalions whose nomenclature links them to the relevant area, ¹⁰ and can be placed in the context of some broader European studies. ¹¹ In ways which no previous local study has been able to do, such academic work has now started to address issues such as the social composition of battalions and the extent to which battalions maintained their local connections as the war progressed. In time, they will allow more nuanced approaches to debates such as those about recruiting patterns. ¹²

Much excellent work has been carried out in the categories above. However, the extent to which any of these studies can be precise about the local connections of any unit has been limited by the nature of the sources on which they are based. Consequently, unit histories, whether of Pals battalions or divisions, generally do nothing more than tell the story of the units concerned. Even where work tries to document the broadly accepted view that the ‘local’ content of most units diminished as the war dragged on, they cannot root this in information on precise place of residence. ¹³ Localized works focus on the dead, often centred on those listed on one or more local memorials. Even if many battalions are covered by a local study, the stories are predominantly those of the dead.

Meanwhile, although there will be cases where Pals battalions recruited most of the available manpower, plenty of other areas which were linked to such a unit would have sent men well beyond the Pals formation. Yet assessing which units men served in, beyond ‘local’ units, has been challenging, again because of the limitations of sources. Academic studies have consciously focused on small selections of battalions. None of these works tells the story of military service in a specific area in its entirety (nor have they sought to). Consequently, what is missing from these local stories is much sense of service in either the Royal Navy or the Royal Flying Corps (and later the Royal Air Force). Men who served with the Royal Engineers, Army Service Corps, Royal Artillery, or other divisional units are usually missing from such studies. That is even the case in divisional histories, which tend to focus on infantry battalions, despite such formations accounting for around one-quarter of those in each division.

Such omissions limit our understanding of the impact of military service on specific areas. If we focus only on units known to be linked to areas, we cannot make an accurate assessment of how far communities were affected by recruitment and casualties (especially spikes in casualty figures) beyond those linked to ‘local’ units. We also gain a distorted view of the war experiences of men who went away and returned having served in non-infantry units with lower fatality rates, in favour of those who served in infantry battalions.

Until recently, nothing could be done to tackle those limitations. In particular, any effort to analyse service from a geographic area was hindered by the lack of availability of soldiers’ home addresses in army records in an easily accessible form. They remain hindered to some extent by problems in accessing naval and flying records. However, the availability of the National Archives’ army service records online ¹⁴ and searchable by place has revolutionized what is possible. These records can be combined with others to give a far more accurate picture of war service in specific areas. Such sources include newspapers, local memorials and listings, and war graves records, which have been favoured by genealogists and amateur local historians but are seldom used by academics. Taken together, these sources mean that we can now use specific geographic areas as the starting point for research, rather than making assumptions about the units in which men served, and that can help us to answer questions about matters such as the social composition and local links of units in a far more sophisticated way than has ever been possible. In particular, the availability of service records online, in a form useable to those analysing local military service, allows such a step change in what can be researched that it should be seen as a ‘new’ methodology. As never before, we now have the opportunity to establish as comprehensive a picture as will ever be possible of how many people from a specific local area served and where they saw service.

Such sources and approaches have of course been used in other areas of history for decades. In the 1960s ‘hard’ social history techniques emerged, based partly on new intellectual approaches which were suspicious of ‘elite’ histories and documentary sources, and sought to gain through statistics a knowledge of the experiences of the downtrodden and voiceless who had not left literary or artistic evidence of their lives. Emerging computer technologies helped to analyse the vast quantity of data used by the new social historians. In the UK these developments manifested themselves in work by figures such as Peter Laslett and Tony Wrigley through bodies such as the Cambridge Group for the History of Population and Social Structure. ¹⁵ Yet such an approach has not been applied, beyond some analysis of enlistment and casualty figures, to the First World War, partly because sources were not available but also probably because issues associated with war did not match social historians’ interests in class, the economy, and everyday life. An exception, published in 1985, when more material had started to become available, was Jay Winter’s The Great War and the British People. It became a classic and remains one of the key texts for study of the UK’s war, but it was almost the first and last of its kind. Within a few years of its publication the cultural turn became dominant in the study of the war, which in some cases became disembodied from the experiences of those who did not leave extensive writings or works of art relating to the war. ¹⁶ Indeed, some such approaches were vulnerable to the criticism levelled at ‘elitist’ history by the new social history.

However, new sources have changed what is possible, and we can now carry out the kind of analysis for the military which was performed for factory and farm workers from the 1960s onwards. We can label such an approach ‘military history from the street’, emphasizing its difference from histories which begin with the units in which men served rather than the areas from where they came. It has the potential to revolutionize the way in which we write the local history of service in the First World War, connecting military experiences with society to open up new areas for statistical analysis, such as the relationship between pre-enlistment occupation and the unit(s) in which men served. The approach thus also offers a new socio-military history of the early twentieth-century United Kingdom. In setting out that approach, this article draws on the methods used for the author’s case study of West Belfast, Belfast Boys, which is to date the only comprehensive example. ¹⁷ It goes beyond that study by offering new data for West Belfast which was not analysed in the book, and then discusses the validity of the approach for other areas. Although central aspects of the narrative of Belfast Boys are unique to Ireland, and some are specific to Belfast, it is argued that the methods used are adaptable to all parts of Britain and Ireland.

II. Military History from the Street

‘Military history from the street’, a new socio-military history approach to the First World War, rests primarily on old sources available in a new way, and combining those with sources which have been underused (or not used at all) by academic historians. At the core of the methodology are the service records of non-commissioned officers and other ranks of the British army. These have been available for public use for many years on microfilm at the National Archives. They are collected in two microfilm series: WO 363, described as ‘service’, and WO 364, labelled ‘pensions’. However, both collections are actually of service records, with the latter being more correctly labelled, as William Spencer has pointed out, as ‘service records derived from pensions claims’. ¹⁸ The reason for their being in two categories was that, after the First World War, any soldier who applied for a pension had his service record removed from others and kept with those of other pension applicants. That was fortuitous, because it meant that more have survived than would otherwise be the case. When the War Office’s Arnside Street repository was bombed during the London Blitz in September 1940, the service records were hit, but those which had been kept separately in relation to pensions (containing the same basic information as the ‘service’ records, plus further information on the pensions claim) were unharmed. Not all of the service records were destroyed, but many were, either through fire or the water used to fight the blaze.

Although the surviving records were later made available on microfilm, it was impossible to do anything other than search for soldiers by name, because they are produced alphabetically. In theory, one could search through all the records for soldiers from a specific area, but a very approximate estimate by this author suggested that if one were able to work 5 days a week full-time on this task, then it would take 27 years to do so, even before one had done anything to write up and communicate findings. Clearly, no academic was going to build a career on such research. Meanwhile, official numerical recruitment returns would be too broad in the area they covered to be of any value for specific local areas. ¹⁹

The situation changed with the publication of the records online. By the spring of 2010 all surviving pensions and service records for NCOs and other ranks were available online. When Belfast Boys was first published, it was still common to hear, among genealogists, references to up to three in four records having been destroyed, and that figure was used as the basis of a very rough estimate of an upper limit of how many West Belfast men might have served. ²⁰ Even with the finalization of the digitization of records, official statements on how many records survived the Luftwaffe’s endeavours are vague and varied. In the first place it is believed that only 20 to 25 per cent of the WO 363 collection has survived, ²¹ but these records are of course supplemented by those from WO 364. One National Archive reference simply talks about ‘more than half’ being destroyed, ²² while another says that ‘About 2 million service records either survived the bombing … or were reconstructed from pensions records’, and says that this constitutes around 40 per cent of the total. ²³ Jay Winter’s figures for overall service suggest that 5,215,162 men served in the British army in 1914–18, of whom 4,948,101 were NCOs and other ranks. ²⁴ Figures obtained from the online provider tell us that there are over 2.9 million individual service records, ²⁵ which would suggest that around 59 per cent of non-officer records survived (although the provider also says that around 60 per cent of records were destroyed ²⁶ ).

However, a sample for men from West Belfast suggests that there is some duplication in the two collections, with around 26.5 per cent of the service records also appearing in the pensions collection. ²⁷ Moreover, there is also duplication within collections. In the service collection, 3014 (27.2 per cent) of the 11,082 records searched were duplicates of others in the collection. In the pensions collection, the figure for duplication was 6.7 per cent (344 of 5129). This means that we can make some estimates about how many individuals there actually are with the online records. A search of the collection shows that there are 1,907,094 records in the service section and 987,589 in the pensions series. ²⁸ The figures for duplication relating to West Belfast records would suggest that this will equate to 1,804,405 individuals, or approximately 36 per cent of non-officers, close to the online provider’s statement that 60 per cent of records were lost. ²⁹

Despite the gaps in the collection, there is still a huge sample of records available to historians. The fact that they are now online revolutionizes the way in which they can be used, allowing research that was never practically possible when they were only accessible alphabetically on microfilm. They can be searched in several different fields, for example by name, service number, regiment, place/year of birth, and, crucially, residence. It is this latter point which offers such potential to academic historians seeking to analyse patterns of service by place. Instead of having no possibility of searching through millions of records on microfilm, a search in WO 363 for a place name such as Manchester, Liverpool, or Hemel Hempstead in the ‘Lived In’ field yields records that are immediately more manageable: 26,830, 9465, and 196, respectively. Some imagination needs to be used regarding search terms where boundaries have changed since 1914–18. For example, the town of Hemel Hempstead now includes the village of Boxmoor (sometimes in the past spelt Boxmore), and searches of such alternatives need to be made to gather in stragglers in the records, but that is easily done. Moreover, not all records have information in the ‘Lived In’ search field (even though the data is usually in the record), and so searching in place of birth is also necessary. When this is all done, it is quite easy to examine and note around 150 of these records in one day’s work, and that makes such a project manageable not over decades but in no more than a couple of years for a larger city, or much less time for smaller areas.

In any such study, boundaries must be clearly drawn. The West Belfast study rested on electoral wards, and that enabled statistics from the 1911 census for the total population of each ward to be used where necessary. A street directory available for the city lists each street and which ward it was in. ³⁰ In possession of such knowledge it was possible to look at all online records which appear with a ‘Belfast’ search, and select only those relating to the west of the city. Meanwhile, other terms for key districts of the city were searched, such as ‘Falls’ and ‘Shankill’ (including its misspelling Shankhill, which historically relates to a parish in County Armagh but was often written on Belfast records). There were problems with the transcribing of street names (such as one example of Crumlin Road in a soldier’s record being rendered as Greenlees Road on the website), which meant that street-name searches had to be carried out with care. However, ‘Belfast’ is hard to misread, and for Belfast Boys searches for that word in the pensions records found 2968 individuals, and in the service records another 6875. When other local terms (including Falls and Shankill, plus less common nomenclature such as Ligoniel and Andersonstown) were also searched for the totals became 5129 pensions records and 11,082 service records. ³¹ Each of these was examined, with the result that 1506 West Belfast individuals were found in the pensions records and 1808 in the service records. ³²

The service records provide the best sources for most statistical analysis of patterns of army service for reasons discussed in the next section. However, if one is seeking to build up as complete a picture of local service as possible, they can and should be supplemented by a range of other sources. Indeed, without these, the majority of individuals will probably be missed: the West Belfast study suggests that over 70 per cent of individuals who were identified ³³ could only be located in other sources (see Table 1). These other sources are principally local memorials, local newspapers, and war graves records, plus a small number of sources which are peculiar to specific parts of Britain and Ireland, or to specific units. Local memorials have long been used by local historians as the basis of their publications. There are three broad types of these: memorials focused on an area, memorials associated with a specific church (usually Anglican), and printed listings. These are usually, though not always, listings of only the dead, but some aimed to include all those who served (such as the Irish Presbyterian roll of honour ³⁴ ). Although these seldom contain a home address or unit details, they can be supplemented with other sources. To obtain unit information the Commonwealth War Graves Commission (CWGC) database can be consulted. ³⁵ Home addresses are more problematic in that they will not always appear in the CWGC database, and are sometimes post-war, but they can often be found in newspapers (see below). The same information can be found in Soldiers Died in the Great War and Officers Died in the Great War. ³⁶ Meanwhile, unless one is trying to link a fatality to a specific part of an area (as was the case in the West Belfast study), it is reasonable to assume that the vast majority of those appearing on a local memorial will not only have some connection with the area but will also have lived there at the time they served. However, where local rolls are used it is crucial to maintain a sense of discipline about them, so as not to over-inflate either enlistment rates or fatalities. As with any reports of a fatality, verification should be made through the CWGC website. ³⁷ Meanwhile, the appearance of a name (without an address) on a local memorial should not result in inclusion in overall statistics unless there is, from some other source, a verified home address or next-of-kin address in the area. That said, where a reliable source provides such an address, one should not feel the need to verify it elsewhere: in most cases, a soldier will only be found in one source, especially where they were not killed, wounded, or a medallist (which is why the service records are such a valuable starting point, because their making did not require anything to happen to a soldier beyond the act of enlisting).

OPEN IN VIEWER

Table 1.

Source origin of individuals in West Belfast study.

Source	Number	Percentage of total
Presbyterian roll only	1763	20.0
Newspapers only	1645	18.7
Service records only	1328	15.1
Pensions records only	1194	13.6
St Michael’s a roll only	573	6.5
CWGC database only	273	3.1
14th RI Rifles roll b only	175	2.0
Wills index only	132	1.5
Troubles-related c only	21	0.2
Private sources d only	19	0.2
United Reformed roll only	10	0.1
Total from one source only	7133	81.1
More than one source e	1665	18.9
Total	8798 f	100

a

Church of Ireland parish covering part of Shankill area.

b

A book, held at the Royal Ulster Rifles Museum, which includes the name, home address, and denomination of those who served in the 14th Royal Irish Rifles. It is believed to be the only nominal roll in existence for any Irish battalion, and means that we have information which is close to being complete for just one battalion.

c

Newspaper reports of deaths during Belfast Troubles of 1920–2, mentioning former military service.

d

Material provided by individuals, mainly in response to appeals in local media.

e

Over half of these are newspapers combined with either the Presbyterian or CWGC records.

f

The figure of 8798 is an upper limit owing to possible duplication of a very small number of records. If all possible duplicates are duplicates, then the number would be 8473 but no lower, although it is unlikely that all would be duplicates.

Newspapers have rarely been used seriously by military historians, perhaps because of suspicions that what is reported publicly must be treated with doubt. There is certainly room for caution. For example, the West Belfast study found 159 deaths reported which could not be verified through the existence of a war grave record (in addition to a total of 2002 confirmed West Belfast fatalities). A small number of these ‘deaths’ were later found to be men who were missing. While the ‘death’ was reported in newspapers, the later news of survival was not, and it is possible that a significant number of the 159 unconfirmed fall into this category. There were also huge variations in the type of information which newspapers would report, as we see from comparing Dublin and Belfast newspapers. The former included casualty listings but no significant information (such as a home address) on NCOs and other ranks. The latter gave much more detail on men of all ranks, ³⁸ admittedly with more detail on officers, but almost always including a home address and the name of the soldier’s next of kin, and sometimes much more – such as workplace or membership of a local organization. Fortunately, most local newspapers were more like Belfast’s than Dublin’s. Unlike most memorials, newspapers not only focused on the dead but included much detail on the wounded, missing, prisoners, and medallists, and sometimes those who just happened to write an interesting letter home. Up until the middle of 1916, the Belfast newspapers freely reported on the battalions in which men were serving, with that information only being absent for the second half of the war. However, where that information is missing it can often be supplemented through reference to war graves records or (for soldiers who survived) the National Archives’ WO 372 series of medal rolls (which rarely include an address but often list a battalion). ³⁹

The CWGC records have already been discussed as a useful source for verifying reports of deaths and also for filling gaps in detail. Each record also contains a field labelled ‘Additional information’ which from 2012 has been searchable online, where previously it was necessary to contact the CWGC to obtain a search of this field. Where the field has been completed, the data usually reveals the name and address of the next of kin, which is almost always a widow or parents.

In addition to those already mentioned, there are many other sources, some only relevant to certain parts of Britain and Ireland, which can be used to supplement the larger sources. For parts of England there is the National Roll of the Great War. ⁴⁰ Ireland is well served with several sources. Ireland’s Memorial Records are now available on CD-ROM and online, ⁴¹ although they rarely include an address and are widely regard as a problematic source. ⁴² A further CD-ROM, compiled by Kiara Gregory, is an index of over 9000 wills of dead soldiers from across Ireland which are kept in the National Archives in Dublin. ⁴³ The IWM’s War Memorials Archive (formerly the UK National Inventory of War Memorials) ⁴⁴ contains an increasingly useful amount of material for all four countries of the UK, while the Irish War Memorials site covers both Northern Ireland and the Republic of Ireland. ⁴⁵ Soldiers Died in the Great War and Officers Died in the Great War never offer an address but do include a general indication (such as the name of a town or city) for place of birth and enlistment and, sometimes, residence.

Beyond these well-established sources there are other ways of pulling together information on local military service. If there are battalion nominal rolls in existence then those can be consulted, although there is a danger that this will distort data towards known ‘local’ battalions. ⁴⁶ Local history societies and appeals through the media can also yield qualitative information on individuals. The extent to which such information will be useful will depend on what the researcher is seeking to establish and the standards set for information to be included. The West Belfast study set a very high standard for inclusion in the database of men from the area. It rested on two principles which aimed to ensure that neither enlistments nor fatalities were overstated. First, any death, whether reported in a newspaper or on a memorial roll, had to be verified by reference to the CWGC’s records, sometimes aided by Soldiers Died to help resolve queries over, for example, the spelling of names. In Belfast Boys, 2002 deaths were verified in this way, with another 159 which were reported in newspapers or found on rolls unverified and clearly stated as such. Second, to be linked to the area, there had to be a record of an address ⁴⁷ of either the serviceman or (so long as there was no evidence for the serviceman’s home being outside the area) his next of kin, for the 1914–18 period, or for the immediate post-war records of the CWGC. It is not adequate, for example, to assume that all those listed on a local memorial roll were still living in that area when they enlisted. This meant that, although a comprehensive search for local rolls was carried out for Belfast Boys, only one parish church roll was of significant value, in addition to the all-Ireland Presbyterian roll. Moreover, it was not adequate to find addresses in the 1911 census because of the very rapid turnover of residence among some sections of the population (see below for further discussion of the value of the census).

Consequently, simply being born in the Shankill or the Falls did not guarantee inclusion. That made the listings in Soldiers Died and Officers Died of no use other than for verifying death or membership of a unit. Similarly, if there had been a Belfast memorial which included names but not addresses of the dead, then that would also have been no use for research unless an address could be found elsewhere. Of course, where a memorial is in a village and the story of an entire village or town is being told, some assumptions can be made, as stated above, but where an analysis of part of a city is being carried out, more specific address information is required. Despite all that, Table 1 shows that, even for research on a city, a very wide range of sources not only can be used but have to be used, if we are seeking as much data as possible.

As Table 1 makes clear, over 80 per cent of records can only be found in one source, and so, if we wish to gather the maximum amount of information possible for a local area, then as many sources as possible must be used. However, because some of the sources involve selectivity in their creation, they must be used with care in statistical analysis, as discussed below.

It is worth noting where the 1911 census can and cannot be used. The recent completion of its publication online for the whole of Ireland ⁴⁸ has excited historians because it does offer huge potential for research, and of course it is also available for the rest of the UK. Parts of the census for Belfast were online well before the all-Ireland collection as part of the BelFam project, ⁴⁹ and from an early stage of work for Belfast Boys the author found the source invaluable for material on individuals. However, the value of the 1911 census for research on 1914–18 is not as great as we might at first imagine. Aside from providing biographical back-stories for select individuals, its main use for Belfast Boys was to supplement information on religious denomination where this was missing from service records (31 per cent of cases). It might be thought that a majority of the gaps could be filled using the census, and every individual with a service record whose denomination was missing was searched for, but it was only possible to find material for a further 5.2 per cent of those searched for (that is, just one-sixth of those who were researched on the census website). That is likely to be because of the relative (compared with now) lack of diversity in names, because of falsification of age at enlistment on service records, and because many people moved on a frequent basis. These factors make it difficult to verify individuals with reasonable certainty.

The question then arises as to whether the census could be used to find other information. Occupation is the most obvious. On the one hand, the census can be used to assess social origins, for example, by recording the father’s occupation in general terms. However, for individuals, a great deal can change in at least three years, and the author does not feel that it is sufficiently rigorous to assume that an occupation in 1911 should be used as the basis of analysis of links between occupation and recruitment in 1914–18. A more reliable method is to use the information we have in service records. Moreover, once one has a record of someone serving in 1914–18 from a relatively reliable source such as WO 363 or 364, the value of looking for that individual in the 1911 census is unclear. One would verify their existence, but unless one feels that there are service records for non-existent soldiers, ⁵⁰ and nobody has ever seriously suggested this, then doing so would have no point. So, while matching to census data might be interesting for, say, the descendants of soldiers, the academic value of gathering such data is limited, though that should not detract from the value of using the census to assess social origins.

III. Analysis of Local Patterns

Once a database or spreadsheet has been created using as wide a range of sources as possible, the data can be analysed. However, for much statistical analysis, only the service records should be used, because they are free from the biases affecting other sources. This is especially the case where politics might affect the recording of information, as it did in Ireland. In the case of newspapers, Protestant and unionist titles tended to focus stories on battalions which were political in their character. Meanwhile, using a Presbyterian roll, when there are only partial rolls for Anglicans and none for Catholics, is problematic. In contrast, one part of the service records is unsifted other than by the random fact of having survived the Blitz, and the pensions part of the collection contains all applicants for pensions.

Yet there are still limits to what these records can reveal, and before establishing what value the sources discussed can have to academics it is important to recognize those. In the first place, very few members of the Royal Navy, the Royal Flying Corps, and the Royal Air Force are to be found in the service records. No information on naval service would be included unless a man had transferred into the army from the navy. As regards flyers, only those members of the RFC who died or were discharged before the RAF was formed would have had their records kept with those of soldiers. ⁵¹ These are large gaps, because 640,237 men served in the navy and 291,175 in the RFC/RAF. ⁵² As regards the army, the files of all the 247,061 men who served as officers are not part of the service records. Even if they enlisted as private soldiers before obtaining their commission, their records were moved to the officer records section. Although correspondence for approximately 217,000 individuals remains intact, ⁵³ that exists only in paper format and cannot be searched by place. Even if it could be, the basic biographical information contained in army forms B196 and B199 was destroyed in the Blitz. ⁵⁴ Of course, there will be a significant number of references to army officers, sailors, and flyers in many of the other sources, such as newspapers and memorials. However, as discussed below, these have their own problems, and it cannot be assumed that they will offer the same kind of coverage of geographic areas as the service records.

In using the sources there are two broad types of analysis which can be carried out. First, statistical analysis of the service records only, with this selection made not only on the basis that they offer the most complete information on individuals, but also because they cover the dead and survivors, and there was no selection (for example by religious denomination) in the way that they were made. Second, all the sources can be combined to offer as complete a picture as possible of service from a specific area, and to allow the production of wide-ranging narratives. Each approach will yield different types of results, and each carries its own health warnings.

The service records contain as much consistent information on soldiers as can be found anywhere. Of course, some of it might be inaccurate (notoriously age) because the form partly relies on information provided on enlistment by the soldier concerned. However, much of the rest of the information was added after enlistment and does not suffer from the same dangers of deception. Initial information provided by the soldier includes name, age, place of birth, occupation, any former military service, marital status, and children. Age is problematic because of men enlisting when they were too young or too old. Details of former military service might also be incorrect because a man might not want to mention that he had, for example, been discharged as unfit for military service on one or more occasions. ⁵⁵ From the point of view of consistency, it is unfortunate that some forms used by the army did not have pre-prepared spaces for the insertion of home address and religious denomination. ⁵⁶ The latter was often added at the top of the second page of the form above next-of-kin details, while the former was compensated for by the inclusion of the name and address of next of kin, and their relationship to the soldier. Where that person was a wife, we can relatively safely assume that the soldier was living with them, and where that person was a parent or parents, we can see at least some connection with a particular area.

Other information about the soldier’s appearance and physique would be added by the attesting officer at enlistment: height, weight, chest measurement, complexion, eye colour, hair colour, and the presence of any distinguishing marks (which often included a description of tattoos). Subsequent information added to the form over the period of a man’s service would be about battalions or other units served in, and any time spent abroad, partly in relation to eligibility for medals. Time in each battalion and time served overseas was recorded to the precise day. Where a man took any classes of instruction while in the army that would be listed, and there would also be a record of any wounds sustained (again, with the precise date stated). If a man was wounded there might be several additional pages inserted to the record detailing the nature of the wound and treatment received, which would, of course, be particularly pertinent for any pension application. Where a man was discharged owing to a wound, such information would be extensive.

Beyond such core information there is sometimes much more. The West Belfast study found cases of extensive correspondence with parents where soldiers enlisted underage and were claimed back. ⁵⁷ There are examples of the problems the War Office faced in tracking down the relatives of a dead soldier in order to give them a ‘Death Penny’. ⁵⁸ There are also many details of mental health problems. Frank Malone was depressed because he believed he had heart disease, and doctors also noted that ‘he is alone and wants to get married to save his soul, as if he died as he is, he fears he would never see the face of God’. He was discharged with melancholia. Meanwhile, Alexander McCartney believed himself to be a wealthy man who was to buy a butcher’s shop on discharge, and regularly said that pigs were outside his hospital waiting to be slaughtered. He was discharged for ‘insanity’ and being ‘delusional’. ⁵⁹

Such information provides unmatched qualitative material for understanding the lives of ordinary soldiers, but there is also much information which can be used quantitatively in many different ways. In the first place, a particularly important part of studying service from a geographic area is assessing in which battalion or other units men served. That helps us to understand how far men were serving in the units believed to be linked to that area, or whether their service was cast further afield. We can thus move beyond infantry-dominated accounts of the war towards an understanding of how many men from an area were serving in, for example, the Royal Engineers or the Army Service Corps.

IV. Levels and Nature of Service

One of the reasons for analysing West Belfast was the dominance of a particular kind of popular narrative about the Protestant Shankill area. That is paralleled by an almost total lack of knowledge of levels of service from the Catholic Falls, other than awareness of men joining the 16th (Irish) Division. The nationalist leader John Redmond was an active recruiter for the latter, out of the belief that such enlistment would help the Irish secure Home Rule. Thus there was a public awareness of two rival volunteer divisions, and little beyond that, and scant detail on either. The dominant Shankill narrative is a loyalist story which has been included on websites and leaflets aimed at tourists, and is also used in a tourist information board provided by Belfast City Council outside the Shankill Memorial Garden. It talks of 760 men who fought in the war, with only 76 coming back. With its focus on men being killed after going over the top on the first day of the battle of the Somme, that account is by implication an infantry-dominated one. ⁶⁰ Yet a street-by-street analysis gives a very different picture. In the first place, ironically for a community which proudly celebrates its role in the British army, this story severely understates both the number of men who served and the number who were killed. However, it overstates the percentage of those who were killed as a proportion of those who served, and inaccurately concentrates on the 36th Division. In fact, not 760 but over 6431 Shankill men have been found to have served. Many more than around 700 were killed: nearly twice that number, 1358. Moreover, there is a sizeable story to be told about service from the (much smaller, around one-third of West Belfast) Falls area, from which 2341 men have been found to have served, with 644 being killed. ⁶¹ Combined with another 26 men whose addresses crossed boundaries between the Falls and the Shankill, ⁶² that makes a total of up to 8798 identified, though because of possible duplication in records the figure might be as low as 8473 (see Table 1, note f).

These figures show that very much larger numbers of men served than are allowed for in a narrative which focuses on the Ulster Division. Yet they do not amount to the final total of those who served, because of the loss of service records. However, we can make estimates, starting from the knowledge that around 64 per cent of the records were lost, and that 36 per cent of records can be found only in service records. We know that in the surviving service/pensions records there were 2522 men whose existence was not documented in any other source. If this represents 36 per cent of the records which can only be found in this way, then we might expect to find as many as another 4115 in the lost records. That would make an upper limit of 13,280 who served. ⁶³ That would constitute roughly up to 66 per cent of the male population of service age. ⁶⁴

For any part of Ireland at this time, the denomination of recruits is a central question, not least because the levels of service according to denomination were much debated at the time, and previous efforts at assessing recruitment according to denomination are problematic. ⁶⁵ It might be hoped that some estimates would be possible using only the service and pensions records because they do not have the same bias as other sources. However, information on religious denomination is not contained in all of these records. A particular problem is that denomination is lacking for some battalions more than others. For example, it is included for 88.2 per cent of the 9th Royal Irish Rifles, compared to only 38.3 per cent of the 6th Connaughts and just 17.9 per cent of the 7th Leinsters. ⁶⁶ However, the disparity between information recorded in different battalions is highly problematic when it is suspected that some battalions began with different religious compositions. There could be many reasons for this. It might be that the staff compiling records for the Leinsters and Connaughts were simply not very diligent. However, it might also be that most or all of those in the Leinsters were Catholics and it was simply not considered necessary to write that down. In contrast, those making records for the Royal Irish Rifles might have felt that, while most/all volunteers at their recruiting offices would be Protestant, it was important to know what type of Protestant they were. Despite that, the 1911 census can be used, subject to the caveats discussed above, ⁶⁷ yet the figures are far from conclusive. They lead to results which put the Catholic proportion of volunteers at anything between 34.5 and 37 per cent in an area where they accounted for 35.9 per cent of the population. ⁶⁸

We can be more clear about the types of units in which men served, beginning with very broad categories using data from the service records about the first destination of recruits. This does not help with many of the problems around the narrative for the Shankill, but it does point out the limitation of infantry-dominated accounts which are so central to much writing about the British army in 1914–18.

As Table 2 shows, around one-third of men were not initially sent to any infantry battalion (regular, service or reserve). The rest were mainly in units such as the Royal Engineers and the Army Service Corps. Consequently, accounts focused on the infantry overlook a significant part of service. Moreover, accounts which focus on one or a few infantry units known to be associated with an area distort the picture of service even further, as Table 3 suggests.

OPEN IN VIEWER

Table 2.

First destination of West Belfast men.

Type	Number	Percentage
Artillery	186	6.1
Army Service Corps	364	11.8
Cavalry	26	0.8
Infantry a	1978	64.0
Other	53	1.7
Royal Army Medical Corps	77	2.5
Royal Engineers	406	13.1
Total	3090	100

a

Includes reservists and a small number of men in Guards regiments, mainly the 1st and 2nd Irish Guards: 18 (0.6 per cent) in total.

Source: WO 363 and WO 364.

OPEN IN VIEWER

Table 3.

Instances of service by West Belfast men.

Unit summary	From service records		From all sources
	No.	%	No.	%
36th (Ulster) Division a	975	36.3	2113	33.8
9th Royal Irish Rifles	237	8.8	537	8.6
Other Belfast infantry	224	8.3	843	13.5
Non-Belfast infantry	127	4.7	308	4.9
Non-infantry b	387	14.4	425	6.8
16th (Irish) Division	350	13.0	528	8.5
6th Connaught Rangers	117	4.4	141	2.3
7th Leinsters	96	3.6	130	2.1
Other infantry	136	5.1	255	4.1
Non-infantry b	1	0.0	2	0.0
Other Irish units	512	19.1	1654	26.5
Other regular/service Irish battalions	174	6.5	582	9.3
10th Division infantry	160	6.0	361	5.8
Royal Irish Rifles 1st and 2nd (Regular) a	156	5.8	587	9.4
Irish Guards	15	0.6	73	1.2
North Irish Horse	7	0.3	48	0.8
10th Division non-infantry	–	–	3	0.0
Other infantry	227	8.5	571	9.1
English regiments	125	4.7	326	5.2
Scottish regiments	98	3.6	237	3.8
Welsh regiments	4	0.1	8	0.1
Other	621	23.1	1380	22.1
Royal Engineers c	195	7.3	226	3.6
Army Service Corps c	150	5.6	197	3.2
Artillery c	136	5.1	218	3.5
Miscellaneous d	117	4.4	394	6.3
Royal Army Medical Corps c	15	0.6	29	0.5
Royal Flying Corps and Royal Air Force	7	0.3	53	0.8
Royal Navy and Mercantile Marine	1	0.0	263	4.2
Total	2685 e	100.0	6246 f	100.0

a

Both the 1st and 2nd Royal Irish Rifles became part of the 36th Division in the later stages of the war, the 1st from February 1918 and the 2nd from November 1917. However, the division had lost its political character by that point, partly because of post-Somme transfers from other units, and partly because of the transfer of units such as the 1st and 2nd RI Rifles. So it would not be accurate to consider service in these two battalions in terms of the popularly held conception of the 36th Division.

b

36th and 16th Division ‘non-infantry’ covers units of the Army Service Corps, Army Veterinary Corps, Machine Gun Corps, Royal Engineers, Royal Field Artillery, Royal Army Medical Corps, and Trench Mortar Battery.

c

Non-10th, 16th or 36th divisions.

d

‘Other’ covers a wide range of units, such as the Army Veterinary Corps and the Labour Corps, not associated with the 10th, 16th or 36th divisions.

e

Represents 2377 individuals. An instance of service has only been recorded where it is possible to be sure precisely which unit the man served in. For example, ‘Royal Engineers’ (non-10th, -16th or -36th Division) includes only those men whose unit was proven to be not in the 36th Division.

f

Represents 5901 individuals. The same method for attributing service was used as described in note e above.

By counting the number of different instances of service in different units, it is possible to see the range of different units in which men served. Table 3 does not include all 8798 men in the database because not all of those have unit-level data attached to them; for example, church memorial rolls commonly stated a regiment but not a battalion. It contains two sets of figures, one from service records only, and one from all sources. The reason for that has already been stated: if we are trying to compare 36th and 16th Division service, the service records are ‘blind’ according to religion or politics. However, the ‘all sources’ data includes material from unionist newspapers (while nationalist ones did not report the war in as much detail) and from Protestant rolls of honour (while Catholic churches did not produce any in Belfast). Thus, in very simple terms, ‘all sources’ should in theory inflate the relative numbers of unionists and Protestants. That at least is what we would expect to happen, and indeed service in the 16th Division does fall significantly as a percentage when ‘all sources’ are added in. However, the ‘all sources’ column also reduces the overall percentage of men who served in the 36th Division, the biggest increases being in the naval service and the two regular battalions of the Royal Irish Rifles. It is difficult to explain why this would be the case, and the main conclusion to draw from it is that looking at percentages of those who served should always be described as being in a broad range. Thus, we might most safely talk of around one-tenth of service taking place in the 16th Division, and around one-third being in the 36th Division.

Table 3 suggests a number of points about service in the British military. As regards the 36th Division, we would expect most men from the West Belfast area to be in the 9th Royal Irish Rifles, formed from the West Belfast Ulster Volunteer Force. However, almost as many men served at some point in another Belfast battalion, mostly the 15th Royal Irish Rifles, which was formed from the North Belfast UVF. That is partly due to ‘North’ Belfast covering a part of the Shankill, and also due to men transferring into the 15th later in the war when the 9th was disbanded. Even more notable, however, is that men were more likely to serve in a non-infantry unit of the division than is suggested by accounts which focus on the infantry. According to service records, nearly 40 per cent of those who served in the division spent some time in a non-infantry unit, principally the Royal Engineers or Army Service Corps. Using all records that figure is much lower (around 20 per cent), but that is probably artificially low because church rolls tended not to report units of, for example, the Royal Engineers, in the same way that they mentioned specific battalions of regiments. Moreover, because casualties would be lower in non-infantry units, men would be less likely to be mentioned in newspapers.

Examining the 16th (Irish) Division, mainly formed from the nationalist community, we can see that, although sizeable numbers served at some point in the 6th Connaughts or 7th Leinsters, which were associated with Belfast, there were also sizeable levels of service in other battalions of that division – nearly 40 per cent of service in that division was outside those two battalions. More widely, there was significant service outside the 16th and 36th divisions: one in five to one in four served in other Irish units, around one in ten in English, Scottish, or (in a small number of cases) Welsh units, and around one in five in other units not associated with either division. Overall, around half of the instances of service (50 to 55 per cent) were outside the two political divisions.

We can gain an even more in-depth understanding of how far men were grouped into specific infantry units by looking at battalion-level data more closely. Table 4 includes data for men drawn from service records only (so as not to increase disproportionately percentages for predominantly Protestant units covered in other sources). The percentage given is a proportion of the 2377 individuals for whom unit-level data exists, and the table includes figures only for those battalions in which more than 30 men served at some point in the war. This method emphasizes army service at the expense of sailors and flyers, and does not therefore indicate service in a battalion as a percentage of overall service. However, it does illustrate patterns of service within the army. On that point, it should be noted that 387 men served at some point in non-infantry formations of the 36th Division, which equates to 16.3 per cent of those men who served in the army. We can conclude from Table 4 that although more men served in the 9th Royal Irish Rifles than in any other infantry battalion, this still accounts for only 10 per cent of those who served. More widely, no other individual battalion accounted for more than 5 per cent of those who served, which illustrates just how broadly service in the infantry was distributed across different battalions.

OPEN IN VIEWER

Table 4.

Instances of service in infantry battalions (excluding reserve battalions) by West Belfast men.

Unit a	Number	Percentage b
9th Royal Irish Rifles	237	10.0
6th Connaught Rangers	117	4.9
15th Royal Irish Rifles	109	4.6
7th Leinsters	96	4.0
2nd Royal Irish Rifles	85	3.6
1st Royal Irish Rifles	76	3.2
6th Royal Irish Rifles	61	2.6
16th Royal Irish Rifles	59	2.5
2nd Royal Inniskilling Fusiliers	56	2.4
14th Royal Irish Rifles	53	2.2
10th Royal Irish Rifles	51	2.1
8th Royal Irish Rifles	31	1.3

a

Includes only battalions where number serving is at least 30.

b

Percentage is of 2377 individuals for whom specific unit/battalion-level data exists, including non-infantry.

Source: WO 363 and WO 364.

One explanation for this spread of service is of course that, although men were likely to be sent to a ‘local’ unit in the early months of the war, later on there were pressures simply to use men wherever they were needed. This can be seen in the extent to which the dominance of recruitment into the 36th and 16th divisions had disappeared by the middle of 1915, as Table 5 shows. It should be noted that because these figures are drawn only from the service records, which are incomplete, actual numbers would be three to four times higher, ⁶⁹ and it is the relative proportions which are the most important indicators.

OPEN IN VIEWER

Table 5.

Recruitment destination per month for West Belfast men.

Unit	Aug–Sep 1914		Oct–Dec 1914		Jan–Mar 1915		Apr–Jun 1915		Jul–Sep 1915		Oct–Dec 1915		Jan–Mar 1916		Apr–Jun 1916		Jul–Sep 1916		Oct–Dec 1916		Jan–Mar 1917		Apr–Jun 1917		Jul–Sep 1917		Oct–Dec 1917		Jan–Mar 1918		Apr–Jun 1918		Jul–Sep 1918		Oct–Nov 1918
	No.	% a	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%
10th Div.	110	16.8	11	2.3	3	0.9	1	0.3	–	–	–	–	–	–	–	–	–	–	–	–	1	1.4	–	–	–	–	–	–	–	–	–	–	–	–	–	–
16th Div.	25	3.8	143	30.3	58	18.1	42	12.3	17	12.5	3	1.9	1	1.1	–	–	–	–	–	–	2	2.8	–	–	–	–	–	–	1	6.7	–	–	–	–	–	–
36th Div.	286	43.7	229	48.5	190	59.4	105	30.8	10	7.4	12	7.7	3	3.2	1	2.0	7	8.0	2	3.6	1	1.4	–	–	–	–	1	4.8	–	–	–	–	–	–	–	–
Other artillery b	20	3.1	31	6.6	7	2.2	2	0.6	5	3.7	9	5.8	3	3.2	1	2.0	1	1.1	3	5.5	2	2.8	4	7.3	3	7.1	1	4.8	–	–	–	–	–	–	1	16.7
Other ASC b	2	0.3	–	–	2	0.6	8	2.3	5	3.7	15	9.7	6	6.3	3	6.0	9	10.2	8	14.5	27	37.5	20	36.4	8	19.0	6	28.6	5	33.3	8	72.7	5	23.8	2	33.3
Other RE b	5	0.8	3	0.6	10	3.1	6	1.8	5	3.7	6	3.9	6	6.3	10	20.0	37	42.0	19	34.5	28	38.9	15	27.3	16	38.1	4	19.0	6	40.0	2	18.2	2	9.5	1	16.7
English, Scottish & Welsh infantry	16	2.4	13	2.8	7	2.2	56	16.4	14	10.3	18	11.6	24	25.3	3	6.0	1	1.1	4	7.3	1	1.4	2	3.6	–	–	–	–	–	–	–	–	1	4.8	–	–
Other Irish units c	50	7.6	8	1.7	4	1.3	5	1.5	4	2.9	5	3.2	5	5.3	1	2.0	3	3.4	–	–	–	–	–	–	1	2.4	1	4.8	–	–	1	9.1	2	9.5	–	–
Other	4	0.6	9	1.9	3	0.9	5	1.5	7	5.1	10	6.5	3	3.2	4	8.0	1	1.1	1	1.8	2	2.8	1	1.8	4	9.5	1	4.8	1	6.7	–	–	2	9.5	2	33.3
Reserves	137	20.9	25	5.3	36	11.3	111	32.6	69	50.7	77	49.7	44	46.3	27	54.0	29	33.0	18	32.7	8	11.1	13	23.6	10	23.8	7	33.3	2	13.3	–	–	9	42.9	–	–
Total	655		472		320		341		136		155		95		50		88		55		72		55		42		21		15	–	11	–	21	–	6	–

a

Some percentage columns might not add up to 100% exactly owing to rounding up/down to one decimal point.

b

A unit other than one attached to the 10th, 16th or 36th divisions.

c

Infantry, North Irish Horse, and Guards.

Source: WO 363 and WO 364.

Until June 1915, the end of the quarter in which the Ulster Division was filled, the dominance of the 36th Division in early recruitment is evident, especially to the end of March 1915. It was followed by the 16th Division, which in the October to December 1914 quarter took nearly one-third of recruits. In addition, figures for reservists should be considered alongside those for the 16th and 36th divisions because the reserve battalions in question were primarily focused on supporting these volunteer divisions.

However, after the two divisions were filled there was a somewhat different pattern, where the steady reduction in the proportion of recruits being sent to a ‘local’ battalion is apparent. While in the first two quarters a clear majority went either to one of the three volunteer divisions or to an associated reserve unit, that began to change in early 1916, with a steady increase in the proportion of men being sent to a unit of the Royal Engineers or Army Service Corps outside the 10th, 16th, or 36th divisions.

These patterns reflect the steady reduction of local links across the British army as a whole. Just as, for example, the 9th Royal Irish Rifles was bolstered by drafts from Norfolk, Hertfordshire, and London after the Somme, ⁷⁰ Belfast men might be sent to any part of the army.

V. Pre-enlistment Employment and Wartime Service

The data on the types of units can be examined in further depth so that we understand not only which units men went into and when, but also which types of men went into which types of unit. Information on the pre-war occupations on recruits allows such analysis. Previous studies, because of the nature of sources used, have not been able to make much comment on the relationship between pre-enlistment occupation and the unit(s) in which men served. Indeed, there has been very little consideration of how far recruiting offices might have paid any attention to occupation when deciding which unit a recruit should enter. Peter Simkins has pointed out the theoretical impact of the individual enlistee’s choice. Timothy Bowman has referred to the attractions of the higher pay in units such as the Army Service Corps. ⁷¹ Meanwhile, there has been some discussion of the relative value of skills to munitions producers or the army, in the context of wider debates on manpower. ⁷² We also have considerable knowledge of relative enlistment rates across different occupations, which is part of a wider debate about how far factors such as age and occupation affected enlistment, and some of this is now focused on specific local studies. ⁷³ In an Irish context David Fitzpatrick has considered the issue partly against the background of the impact of pre-war militarism in Ireland. He has examined the impact of occupation on propensity to enlist, perhaps most importantly finding that ‘the relationship between enlistment and insecurity of employment was inverse rather than direct’, with those who had the strongest economic interests for not enlisting instead doing the opposite. ⁷⁴ However, beyond brief discussion of the fact that, from late 1914, pioneer battalions sought men who had experience of digging or tunnelling, ⁷⁵ there has been very little consideration of how far men with specific skills might be used in specific ways by the army and whether the skills needs of the British army were met through the distribution of skilled men into suitable units. Typical accounts of the formation of the new armies either pay little attention to the backgrounds of recruits, ⁷⁶ or put forward the popular view that infantry battalions contained men from a very wide variety of backgrounds, even in some Pals battalions, ⁷⁷ although there is obviously much work showing how Pals battalions and some other units were more narrowly focused in social terms. ⁷⁸

The impact of occupation on the type of unit a man joined can be looked at in different ways. In the first instance, patterns for an area across the entire war can be analysed. In the case of West Belfast, the hypothesis investigated (by this author and a colleague ⁷⁹ ) was whether assignment to specific types of military units was systematically influenced by the pre-war occupation of the recruit. To test this out the data from 2675 recruits was categorized by unit and occupation. Unit categories were: artillery, ⁸⁰ Royal Engineers, infantry, ⁸¹ Army Service Corps, and other. ⁸² Occupation categories were: animal related, ⁸³ commerce, ⁸⁴ other skill, ⁸⁵ skilled construction, ⁸⁶ skilled engineering, ⁸⁷ skilled manufacturing, ⁸⁸ skilled trades, ⁸⁹ unskilled, ⁹⁰ and white collar. ⁹¹

The statistical analysis conducted was a multinomial logistic regression. ⁹² The dependent variable was the assigned unit and the predictor variable was the category of pre-war occupation. Logistic regression computes a series of odds ratios (ORs). These compare the odds of being assigned to one unit relative to a reference unit (‘infantry’ in this case) for each pre-war occupation relative to a reference occupation (in this case, ‘unskilled’). The overall logistic regression shows that the unit to which a recruit was assigned in general was strongly and reliably influenced by occupation prior to recruitment into military service (likelihood ratio test statistic=308.2, df=32, p<0.0001). Of more interest, however is the specific pattern of ORs, as set out in Table 6.

OPEN IN VIEWER

Table 6.

Destinations of recruits by unit and occupation, relative to the infantry and unskilled workers from West Belfast.

Unit type a		B	Std error	Wald	df	Sig.	Exp(B)	95% CI for Exp(B)
								Lower bound	Upper bound
Artillery	Intercept	−2.773	.127	477.514	1	.000
	Animal related	1.460	.327	19.978	1	.000	4.308	2.271	8.173
	Commerce	.644	.490	1.731	1	.188	1.905	.729	4.975
	Other skill	.758	.360	4.435	1	.035	2.133	1.054	4.318
	Skilled construction	−.318	1.030	.096	1	.757	.727	.097	5.479
	Skilled engineering	.557	.329	2.859	1	.091	1.745	.915	3.329
	Skilled manufacturing	−.066	.323	.041	1	.839	.937	.497	1.764
	Skilled trades	.146	.441	.109	1	.742	1.157	.487	2.747
	White collar	.495	.540	.841	1	.359	1.641	.569	4.730
Army Service Corps	Intercept	−2.118	.094	508.578	1	.000
	Animal related	1.856	.230	64.910	1	.000	6.396	4.072	10.046
	Commerce	1.019	.323	9.987	1	.002	2.772	1.473	5.216
	Other skill	1.202	.236	26.028	1	.000	3.326	2.096	5.277
	Skilled construction	.636	.504	1.593	1	.207	1.890	.703	5.077
	Skilled engineering	1.388	.192	52.370	1	.000	4.006	2.751	5.834
	Skilled manufacturing	.666	.186	12.854	1	.000	1.947	1.353	2.802
	Skilled trades	.407	.296	1.895	1	.169	1.503	.841	2.684
	White collar	1.345	.300	20.092	1	.000	3.838	2.131	6.910
Other	Intercept	−2.835	.131	470.710	1	.000
	Animal related	1.186	.369	10.327	1	.001	3.275	1.589	6.754
	Commerce	.484	.539	.804	1	.370	1.622	.564	4.668
	Other skill	.127	.480	.070	1	.791	1.135	.443	2.909
	Skilled construction	−.256	1.031	.062	1	.804	.774	.103	5.838
	Skilled engineering	.774	.312	6.133	1	.013	2.168	1.175	3.999
	Skilled manufacturing	.077	.314	.060	1	.806	1.080	.583	2.000
	Skilled trades	.613	.374	2.684	1	.101	1.847	.886	3.848
	White collar	1.117	.431	6.729	1	.009	3.057	1.314	7.112
Royal Engineers	Intercept	−2.253	.100	509.716	1	.000
	Animal related	1.192	.291	16.762	1	.000	3.293	1.861	5.826
	Commerce	.124	.483	.066	1	.797	1.133	.439	2.922
	Other skill	1.026	.262	15.321	1	.000	2.791	1.669	4.665
	Skilled construction	2.253	.318	50.312	1	.000	9.514	5.105	17.729
	Skilled engineering	1.711	.186	84.446	1	.000	5.535	3.843	7.973
	Skilled manufacturing	1.041	.177	34.675	1	.000	2.831	2.002	4.003
	Skilled trades	2.068	.191	117.175	1	.000	7.909	5.439	11.501
	White collar	1.228	.327	14.095	1	.000	3.415	1.799	6.484

a

The reference category is infantry.

Note: the table displays the parameter estimates from a multinomial logistic regression. Exp(B) = the odds ratios referred to in the text.

Source: WO 363 and WO 364.

Key points which emerge from the data are that the odds of being posted to the Royal Engineers relative to the infantry were more than three times greater for those who had worked in animal-related professions than for unskilled workers (OR=3.29, 95% CIs=1.861, 5.826), and this effect was highly significant statistically (p<0.0001). The impact of having worked in an animal-related occupation can be seen even more strikingly elsewhere. Such men were more than four times more likely to go into the artillery relative to the infantry than were unskilled workers (OR=4.31, 95% CIs=2.271, 8.173), and more than six times more likely to go into the Army Service Corps (OR=6.39, 95% CIs=4.072, 10.046). In both cases, this effect was again highly significant statistically (p<0.0001).

Meanwhile, the Royal Engineers were more than nine times more likely relative to the infantry to recruit skilled construction workers (OR=9.51, 95% CIs=5.105, 17.729), more than seven times more likely to recruit skilled tradesmen (OR=7.91, 95% CIs=5.439, 11.501), and more than five times more likely to recruit skilled engineers (OR=5.54, 95% CIs=3.843, 7.973). In both cases, this effect was again highly significant statistically (p<0.0001). By contrast, the odds of being posted to the Royal Engineers relative to the infantry was not significantly higher for those who had worked in commercial professions compared with unskilled workers (OR=1.13, 95% CIs=0.439, 2.922).

In sum, the results show that those who worked in animal-related occupations and other skills tended to go into the artillery. Meanwhile, every occupation other than construction and trades was most likely to go into the Army Service Corps, every skilled occupation (not ‘commerce’) went into the engineers, and the other units received relatively more postings from animal-related, engineering, and white-collar backgrounds. This seems to reflect a very rational pattern of postings, utilizing pre-war work skills as one basis for the recruits’ destination units. Whether or not this was unique to West Belfast, Belfast, Ireland, or the UK more widely would of course require further local study. Moreover, we cannot be clear exactly why this was the case. It is possible that a decisive factor was that an enlistee with a particular skill could earn more in units which could utilize that skill than he could in the infantry. ⁹³ However, the surviving records rarely indicate any preference on the part of enlistees, even though some forms contained a space for that to be indicated. But the data which does survive does open up significant new territory for academic analysis.

VI. Local Narratives

Beyond statistical analysis, the sources discussed above can also be used to inform narratives of local areas. The statistics will be crucial for determining which units are most important to include, and the individual records will also reveal detailed personal stories, as already discussed. However, it is only in conjunction with other sources that these can be used to construct a narrative which reveals the full richness of a local area’s service in the First World War. In the first place, battalion war diaries are an essential way of matching a soldier’s dates of service with specific locations at the front. With such war diaries containing day-by-day (often hour-by-hour and sometimes minute-by-minute) accounts of where a battalion was and what it was doing, it becomes possible to flesh out what a group of men, who might have arrived in France in October 1915 and stayed there for much of the war, would actually have been doing. There are even cases where the death or wounding of a soldier can be linked to a specific incident recorded in the war diary, even if the man is not named in the latter. For example, in the West Belfast study it was possible to match the death of a man (Private Robert Fitzgerald, 6th Royal Irish Rifles) in Salonika to an isolated training accident recorded in the war diary, even though no report of the nature of the death was found elsewhere. ⁹⁴ On only a slightly larger scale, the names and addresses of seven men killed on the night of 5–6 June 1916 could be matched by combining newspaper reports and the war diary. This showed that six of the men killed in this one incident of enemy shellfire had all lived very close to each other in the Shankill area of Belfast. The incident contributes to a more nuanced understanding of Shankill fatalities than has previously been the case because of an excessive focus on 1 and 2 July 1916.

Such cases point to the value of newspapers as a source. In the first place they carry very detailed information of casualties and medallists. Military historians can be rightly sceptical of material in newspapers because they were so often used as propaganda, but they do contain information that can help to challenge some of the narratives of the war which have a popular hold today. In the case of West Belfast, for example, there are strong narratives around the battalions of the 36th (Ulster) Division, in particular the 9th Royal Irish Rifles formed from the West Belfast Ulster Volunteer Force. In recent years, during the Northern Ireland peace process, a new narrative has emerged around the Irish nationalists who joined the 16th (Irish) Division at the behest of the nationalist leader John Redmond. Many of those from West Belfast went into the 6th Connaught Rangers. However, both narratives share a core strand which sees volunteers joining for political reasons, and going on to serve both on the Somme (thought at different stages in the battle) and then side by side at Messines in June 1917. Beyond this, little else about the war is remembered. Newspapers offer a contrast. For all that newspaper editorials might focus on the political divisions, the individual stories contained in casualty lists offer an alternative implicit and subversive narrative. Even where stories were being produced for propaganda reasons, they illustrate the diversity of West Belfast men’s military experiences, from the outbreak of the war to its very end, and in every theatre in which the war was fought. Thus the war goes beyond the Western Front to draw in Gallipoli, plus isolated but prominent examples of service in Africa, on the Eastern Front, and, with the Royal Navy, as far away as the Falkland Islands. ⁹⁵ The war is also extended chronologically as well as geographically, with the initial retreat at Mons and every battle through to the eventual victory. Consequently, for any part of Britain and Ireland where stories of one or more local units dominate the narrative, the scope of the war can be reconceived as contemporaries saw it. Newspapers can inject into a simplistic narrative a view of the war that would have been held by local people at the time but has long been forgotten in the popularizing and ritualizing which have accompanied remembrance.

VII. Conclusions

One case study can only scratch the surface of the academic value of the service/pensions records, supplemented with other sources. Some of the value of that study relates to the very specific issues involved in the service of men from the island of Ireland in the British army, particularly denominational segregation in battalions. However, it also points to possibilities which go beyond the traditional outputs of military historians, towards histories which root men in the streets in which they lived and in their places of work. In other words, newly available sources combined with old sources used in new ways offer the prospect of putting the social into military history to create a new sub-discipline: socio-military history, or ‘military history from the street’.

There is scope for the further development of such an approach. In particular, data could be used to explore differences between volunteers and conscripts. This was not possible for West Belfast because there was no conscription in Ireland. However, for Great Britain, such data could help remedy the problem that existing literature is focused on volunteers. This would allow quantitative evidence to be brought to bear on debates such as the propensity of different occupations to enlist, with that information set against a range of other factors such as age, marital status, and whether or not a man had served in the military before.

Before we can analyse service in local areas in as ‘complete’ a manner as possible, further records will need to be digitized and made searchable by place: naval and RFC/RAF records and those of army officers. But even without that, there is much that we can learn about the ordinary soldier. For this to happen, we need of course to have more local studies carried out to offer comparative perspectives and perhaps, in time, to build up a national database. If that is to be done effectively then there needs to be a dialogue between academic historians, amateur local historians, ⁹⁶ and genealogists. All three will have different reasons for gathering information but they will largely be compiling the same type of data. While some academics can take a rather elitist view of the work done by so-called ‘amateurs’, anybody who has engaged with local groups and individual projects will be well aware of not only the energy which they apply to tasks, but also the very high standards on which much ‘amateur’ work rests. However, if data is to be compared across projects, it is important that it has been recorded on the basis of the same categories and terminology (especially when it comes to abbreviations). For that reason, the academic advice being given to the IWM’s crowd-sourcing project, Lives of the First World War, ⁹⁷ will be crucial in setting standards which researchers can follow. With such standards established and applied, we will be in a position to take our understanding of the place of the First World War in British and Irish society down to a street level.

Perhaps surprisingly, such a ‘low’ approach might also speak to debates in ‘high’ politics and military strategy which have traditionally been focused on questions about the learning and competence (or otherwise) of the upper echelons of officers, and the agendas of government ministers. Yet wars are fought by the men on the ground, and the success or otherwise of battle plans must be influenced in some way by the nature of the troops available to the generals. How far the infantry would be capable of achieving the goals set for it might be affected significantly by the skills of its men. Only by understanding the nature of those men, in different units and at different stages of the war, can we really understand how far they could be expected to be effective soldiers. The next dimension of debates on learning curves and the competences of generals might well be found among the ranks.