By Cassie Watson; posted 11 December 2019.
As the current election campaign draws to a close amid increasingly shrill claims and counterclaims, I am reminded of a saying that, while common today, appears to have originated around the time of a much earlier election campaign. In early July 1892 remarks made by “Mr Balfour” — presumably Arthur Balfour MP (1848–1930), then Leader of the House of Commons — were cited by fellow Tories in relation to the three categories of untruths alleged to be inherent in Gladstonian speeches about Irish Home Rule: “lies, damned lies, and statistics.” While the possibility of (another) minority government is one route to pursue regarding this observation, I am more interested in the statistics part of Balfour’s quotation, as the collection, analysis and interpretation of historical data are among the main methodologies utilised by criminal justice historians. The saying itself is now understood to mean that numbers have the power to bolster weak arguments, but are statistics really no better than lies?
As Clive Emsley has noted, “Many of the key questions, certainly many of the most popular questions about crime, are quantitative.” Historians want to know how much crime there was, who committed it, what happened to them and how this changed over time; yet, “the statistical evidence of crime is fraught with dangers and difficulties.” Historians of the eighteenth century, and earlier periods, have to compile their own figures by counting documents in archives, a process that has its own pitfalls; but since 1805 central government has made the task easier by compiling annual figures of indictable offences, joined in 1836 by commitments to prison. The government also published increasing amounts of information relating to coroners and inquests, criminal lunatics, death sentences and reprieves, magistrates, the police (who produced a large amount of data in their own right), prosecution costs, and probably much else besides; these are simply the types of data that I have found particularly useful in recent years. Irrespective of longstanding doubts about the accuracy of these figures, however, especially with regard to their uniformity, the ‘dark figure’ of unreported or unrecorded crime and bureaucratic “rationing” of criminal prosecutions, we can agree with Emsley and R. S. Sindall that “Their value lies in forming the basis of the picture of crime which powerful influences in nineteenth-century society thought confronted them.”
But statistics can do much more besides, for the patterns they reveal provide historians with clear pathways into the study of a wide range of criminal behaviours, as well as providing insights into the workings of the criminal justice system. Statistics compiled by government or other bodies should not be the endpoint of research, but a starting point. Some examples should serve to demonstrate what I mean.
Several years ago I stumbled across John Powell’s publication, for the London Statistical Society, of a range of national statistics compiled from government sources in the wake of the economic hardship that followed the end of the Napoleonic Wars. A simple table organised by assize circuits (p.14) includes both a ratio of county population for England and Wales and the total number of county magistrates. This gives an instant picture of how crime might have been tackled differently in various counties. While Staffordshire had a population approximately the same as Gloucestershire, it had fewer than half the magistrates. Tiny Anglesey, Merionethshire and Radnorshire, by contrast, whose populations were apparently too small to signify, seem to have been comparatively well supplied with JPs. At a time when crime control still fell largely on the shoulders of the local population and parish constables, the easy availability of a magistrate might well impact recorded levels of crimes such as assault. While not all magistrates were active, in this case it is extremely interesting to note that the Crime and Punishment database assembled by the National Library of Wales includes nearly 3,600 cases of assault, a crime that has not yet been sufficiently explored by historians.
More recently, I was investigating pre-war murder rates, and was struck by two things: the relative infrequency of homicide in general and poisoning in particular, in comparison to Victorian levels. On the left, figures for 1913 show 178 murders, of which 37.6% involved infants under the age of one year. On the right, figures for 1930–1938 show that the absolute number of murders recorded had dropped significantly, but quite a few cases involved more than one victim. However, infants were a smaller proportion of all victims, falling from 27.7% in 1930 to 17.0% in 1938. This is interesting, given the intervening population increases: after World War One infants were less often the victims of murder; which was itself slightly less frequent. The absolute numbers are less important than the trends, but can be further investigated using newspapers and coroners’ returns to identify cases of dead infants in which no official determination of homicide was made.
The table below highlights an important distinction between the two sets of data: the Infanticide Act of 1922 established a separate offence applicable only to new mothers. (This table also includes incest, a much more hidden offence for which no official figures were collected before 1909.) There is another difference, too: Table A (1900–1938) enumerates persons for trial, not the number of cases or victims. The discrepancy in the figures for 1938 — 100 cases of murder but only 66 individuals tried — indicates that a significant number of alleged perpetrators did not enter the criminal justice system, diverted by suicide (which was quite common among murderers in 1938) or committal to an asylum as a criminal lunatic, or because no arrest was made during the year the data was collected; however in a few cases more than one suspect was tried for a specific murder. All of these subjects form starting points for potentially fruitful comparative or longitudinal research projects, undertaken at national or local level.
The Home Office statistics do not state cause of death, merely the total numbers of murders and suicides. However, coroners and the Registrar General collected such information, and figures released by the latter for 1932 showed the highest-ever suicide rate then recorded: 5,743 people had taken their own lives in England and Wales, in a year when 124 victims of murder were known to the police. Douglas Kerr, a medico-legal expert in Edinburgh, used this data to identify all the deaths from poisoning; of which 2,584 were suicides (69.9% caused by coal gas) and 289 were accidents (69.2% due to coal gas). Apart from the two obvious points of interest here, there is a third: why didn’t Kerr list any poison murders for 1932?
The 1930s was notable as a decade in which three women were executed for poison murders (in 1934 and 1936), but the twentieth century on the whole was not one for which executions reveal a penchant for poison, used by only 3% of executed murderers. Manslaughter convictions might be more illuminating, but it is likely that many poisonings were not identified as criminal, and that criminal poisoning was actually on the wane.
This leads to my final example, which shows what statistics can and cannot reveal. In 1928 Sir William Willcox, a renowned toxicologist and Home Office Analyst, noted that advances in tests for arsenic continued to be important because of its use in therapeutic compounds like salvarsan. He mentioned Beatrice Pace, whose sensational trial for husband murder ended in acquittal in July 1928; and also the case of Jessie Llewellyn, a 53-year-old invalid who died in June 1928, allegedly poisoned by her husband. Clearly influenced by the Pace trial, this case is interesting for what it tells us about popular suspicions of poison in an era when murder by poison was less common than it had once been. This “poison death riddle” had confused the local doctor but Willcox correctly identified the source of the arsenic in her body as coming from her seafood diet. The inquest was counted in official statistics that recorded Mrs Llewellyn’s death as arising from natural causes, but the newspapers provide the real historical detail. Ultimately, statistics can be useful tools to think with, whether they are accurate or not, and can be very helpful in identifying important information to collect, analyse and interpret.
Home Office publications were retrieved from ProQuest and snippets from them are reproduced under the Terms and Conditions for Authorized Users.
John Powell, Statistical illustrations of the Territorial Extent and Population; Commerce, Taxation, Consumption, Insolvency, Pauperism, and Crime, of the British Empire (London: J. Miller, 1825). Retrieved from the Open Library.
“Grave Evidence for Jury in New Arsenic Case,” The People, 15 July 1928, p. 1. Retrieved from the British Newspaper Archive and reproduced by permission. Image © The British Library Board. All rights reserved.
 Pall Mall Gazette, 6 July 1892, p. 1; Coventry Evening Telegraph, 8 July 1892, p. 3; The Scotsman, 31 August 1892, p. 10.
 Clive Emsley, Crime and Society in England 1750–1900, 5th edn (Abingdon: Routledge, 2018), p. 21.
 Shannon McSheffrey, “Detective Fiction in the Archives: Court Records and the Uses of Law in Late Medieval England,” History Workshop Journal 65 (2008): 65-78.
 R. S. Sindall, “The Criminal Statistics of Nineteenth-Century Cities: A New Approach,” Urban History 13 (1986): 28-36.
 Howard Taylor, “Rationing Crime: The Political Economy of Criminal Statistics Since the 1850s,” The Economic History Review 51 (1998): 569-590; Howard Taylor, “The Politics of the Rising Crime Statistics of England and Wales, 1914–1960,” Crime, History & Societies 2 (1998): 5-28; Robert M. Morris, “‛Lies, Damned Lies and Criminal Statistics’: Reinterpreting the Criminal Statistics in England and Wales,” Crime, History & Societies 5 (2001): 111-127.
 Sindall, “The Criminal Statistics,” p. 35; Emsley, Crime and Society, p. 48.
 John Powell, Statistical illustrations of the Territorial Extent and Population; Commerce, Taxation, Consumption, Insolvency, Pauperism, and Crime, of the British Empire (London: J. Miller, 1825).
 Their populations were approximately, at the 1821 Census, 45,000, 34,000 and 23,000 respectively. There is a serious error in Powell’s book, which states (p.11) the population of Merionethshire in 1821 to be 233,900!
 Judicial Statistics, England and Wales, 1913, Part I–Criminal Statistics, Cd. 7767, 7807, Vol. LXXXII.1, p. 87.
 Home Office, Criminal Statistics England and Wales 1938, Cmd. 6167, Vol. XI.251, p. xviii.
 12 & 13 Geo V c.18, An Act to provide that a woman who wilfully causes the death of her newly-born child may, under certain conditions, be convicted of infanticide. This statute created a new offence equivalent to manslaughter, and applied to any woman found to have deliberately killed her newborn child while the balance of her mind was disturbed because of giving birth.
 Shani D’Cruze, Everyday Violence in Britain, 1850–1950 (Harlow: Pearson Education, 2000), pp. 130, 131.
 Home Office, Criminal Statistics England and Wales 1938, p. 181 indicates that 39 of 55 alleged murderers committed suicide at the time of the murder.
 Gloucester Citizen, 19 December 1933, p. 11.
 Home Office, Criminal Statistics England and Wales 1932, Cmd. 4608, Vol. XXVI.599, pp. x-xi.
 Douglas Kerr, Forensic Medicine: A Text-Book for Students and a Guide for the Practitioner, 2nd edn (London: A. & C. Black, 1936), p. 212.
 Peter Wilson, Twentieth Century Hangings (London: Blackie & Co., 2002), p. 639
 John Carter Wood, “The Most Remarkable Woman in England”: Poison, Celebrity and the Trials of Beatrice Pace (Manchester: Manchester University Press, 2012).
 Sir William Willcox, “Acute Arsenical Poisoning,” Transactions of the Medico-Legal Society 23 (1928-1929), pp. 153-155.
 Ibid., p. 154; Western Mail, 14 July 1928, p. 11.