6 The rise of statistical thinking in the 19 th century
6.2 Establishment of the infrastructures of statistics
the infrastructures of statistics
Westergaard (ibid.) called 1853–1888 the period of statistical congresses because during that period a series of international statistical congresses were organized.
The importance and status of statistics was corroborated in successive meetings.
Westergaard (ibid.) claims that the initiative leading to the establishment of the international statistical congress was principally due to Quetelet51. The fi rst
con-51 Adolphe Quetelet (1796–1874) was a Belgian scientist. He received his fi rst doctorate in 1819 and after receiving this doctorate he taught mathematics for a while. In 1823, he went to Paris to study astronomy. Aside of astronomy he learned the theory of probability under Fourier and Laplace. In Belgium he fi rst worked at an observatory but later he was appointed as the director of statistical bureau.
gress was held in Brussels in 1853 with Quetelet as main organizer and chairman.
In the following 23 years, eight other congresses took place.
The activity around statistics was remarkable already before statistical Con-gresses, and throughout Europe, different bodies for statistics were established, such as statistical institutions and statistical societies. Especially the impact of the statistical societies was signifi cant for the development of modern statistics and the professional principles of statistical work.
6.2.1 First statistical societies
Willcox (1934) listed fi fteen statistical societies founded between 1834 and 1844.
There existed some societies already before that period, such as the French Sta-tistical society and those of Württemberg, Marseilles and Saxony. Westergaard (ibid.) commented that “Everybody seemed to have got statistics on the brain!”
The societies were mainly founded as citizens’ organizations outside statistical institutions. The most prominent motive seems to have been the interest in social questions (see Westergaard 1932 and Desrosiéres 1998). It is noteworthy that the foundations of the professional character of statistics and statistical work were laid by the fi rst statistical societies.
Statistical societies in England
According to Westergaard (ibid.), the most noticeable development in statistical societies took place in England because in a short period more societies were founded there than anywhere else. The fi rst statistical society in England was founded in Manchester in 183352. Only few of the founding members were statisticians in the modern sense. The founders were partly driven by alarm over the acute social conditions in Manchester. The population of Manchester had grown by 45 per cent between the censuses of 1821 and 1831 as a consequence of rapid industrialization. This, in turn, caused an expansion in employment and it brought acute housing problems and diseases in its train. The objects of the Society were stated as “The collection of facts illustrative of the condition of Society and the discussion of subjects of Social and Political Economy, totally excluding party politics”53.
The Manchester Statistical Society was a pioneering organization also in an-other respect. It was the fi rst institute in Britain to systematically study social problems and to collect statistics for social purposes. In 1834, it carried out the fi rst house-to-house social survey in England. The survey was composed of interviews of 4,102 families of working men in Manchester. One of the fi rst published reports of social surveys was that of Heywood’s Report of an Enquiry, conducted House to House, into the state of 176 Families in Miles Platting, within the borough of Manchester, in 1837 (Heywood 1838). Later, the members of the society carried out several other surveys, including a survey of the state of educa-tion in Manchester and the surrounding boroughs.
52 Statistical Society of Manchester still exists and is working actively.
53 The early history of the Society has been well documented by Thomas S. Ashton (1934).
The best known statistical society in Great Britain is the London Statistical Society (LSS), founded in 1834. The society grew rapidly, and in 1838 it started to publish a journal which is now known as the Journal of the Royal Statistical Society (JRSS). In 1887, the society was granted a royal charter and the Statisti-cal Society of London became the Royal StatistiStatisti-cal Society54.
Soon after the Statistical Society of London was founded, similar societies were founded in larger industrial cities in Great Britain. This rapidly growing interest in statistical matters was associated with the British Statistical Move-ment, which was actively taking part in the work of the statistical societies. As an offspring of this activity, Charles Booth (1840–1916) devoted his fortune to surveying poverty in London. Among other things, he carried out a survey. Its results were published in a book entitled “Life and Labour of the People in Lon-don” (Booth 1889–1903). Benjamin Seebohm Rowntree (1871–1954) adopted Booth’s method to study other English towns and compared them to London.
The most famous of these is a comprehensive survey that he carried out on the living conditions of the poor in York. It was a complete enumeration dur-ing which investigators visited every workdur-ing class home (see Rowntree 1901).
Arthur Bowley has been regarded as a successor of this Statistical Movement (see Chapter 8).
Other famous statistical societies
One of the most famous statistical societies was founded in the Kingdom of Saxony in 1831. Already in the same year, the society published the fi rst issue of its journal, Mittheilungen des statistischen Vereins für das Königreich Sachsen.
The initiative to form the association came from the government and it got its mandate from the king of Saxony. In addition, a considerable number of private persons, “patriotic citizens”, interested in statistics assisted the Society. In 1850, it was converted into a public institution of Saxony.
The American Statis tical Association was founded in Boston in 1839. It was originally called the American Statistical Society, but the name was changed to the American Statistical Association (ASA) at its fi rst annual meeting in 1840.
In 1888, the society started a new publication that later became the Journal of the American Statistical Association (JASA).
6.2.2 Statistical institutes
During the Era of Enthusiasm, offi cial statistics made considerable steps for-ward, not only as to the quantity of collected data but also as to its quality.
This fact, in combination with the work of statistical societies, made the avalanche of printed numbers possible (Hacking 1990).
In the Era of Enthusiasm, there was also remarkable activity in Germany.
The Tariff Union was founded in 1833. It re quired regular censuses in all the German States that joined the Union, because the income from the tariffs
54 The history of the society has been documented in great detail for example by Mouat (1885) and Hill (1984).
was distributed according to the number of inhabitants. Therefore, there were regular triennial enumerations until 1866.
Statistical activity in German states was already notable before the Tar-iff Union. The fi rst statistical bureau in Germany, Königlich Preußische Sta-tistische Bureau, was established in 1805 in Prussia. In 1808, the Statistisch Topografi sches Bureau was established in the Kingdom of Bayern and in the next year it published its fi rst yearbook. In 1807, it had already published a statisti-cal atlas (Statistische Darstellung der Königlich-Baierischen Staaten). In 1821, in the Kingdom of Württemberg, a statistical bureau, Statistisch Topografi sches Bu-reau des Königreichs Württemberg, was established. In 1826, it published its fi rst yearbook (Württenbergische Jahrbücher für Vaterländische Geschichte, Geographie, Statistik und Topographie).
According to Westergaard (1932), noticeable progress was also made in Eng-land, where a statistical department was added to the Board of Trade in 1833.
Equally important was the establishment of civil registration of vital statistics in 1837 for the registration of marriages, births, and deaths.
In France, the Bureau de la Statistique generale was re-establishment in 1833. I had been suppressed in 1812. The bureau had charge of several impor tant subjects, such as population, fi nance, foreign trade and prices. But there was no decided centralisation, various branches being treated separate-ly. A statis tical service was created in 1844 and the following years in the Ministry for Travaux public (Westergaard 1932).
In Belgium, a Statistical Commission was organized in 1841 with the ob-ject of controlling the various branches of statis tics, and Adolphe Quetelet became its president. The most signifi cant event was the census of 1846, em-bracing population as well as agriculture and industry. The industrial census had a detailed classifi cation of profes sions. The report contains the number of work-ing men, their wages, engine horsepower, the num ber of looms and other utensils employed. The whole census was generally looked upon as a very important step forward and it was held as an example of a perfect census by other countries.
In Russia, the development was slightly different from the other countries in Europe. In 1864, provincial and district Zemstvo institutions were created in 33 districts of Russia. They were controlled by the Ministry of the Interior and the respective governors. Many Zemstvos started to organize their own local statistics for their needs, and by the end of the 19th century, 25 out of the 33 provincial Zemstvos had statistical bodies. Kaufman (1918) describes in detail the organisation and statistical activity of zemstvo offi ces.
Zemstvo administrators also carried out statistical surveys. Mespoulet (2002) argues that the quantity and diversity of statistical data needed by Zemstvo administrators stimulated methodological innovations in the fi eld and infl u-enced the rise and development of sampling in Russia at the end of the 19th and beginning of the 20th century. Mespoulet (ibid.) also argues that Kovalevsky’s mathematical treatment on sampling theory and stratifi ed sampling, published in 1924, is a synthesis of the Zemstvo statisticians’ sampling practices and Rus-sian academic statisticians’ theoretical work.
6.2.3 The International Statistical Congresses
An important step forward was the fi rst International Statistical Congress held in 1853. The chief object of the congress was a practical one, namely to promote the organization of offi cial statistics and to unify the reports from various sta-tistical institutions so as to make the documents comparable, and a cheap and easy exchange of statistical publications was recom mended. The tendency in the
“Congress Period” was chiefl y to establish central statistical bureaus.
The program of the fi rst conference covered the whole fi eld of offi cial statis-tics of that time, and the result of the meetings was a list of more or less detailed resolutions. As the aim was practical, there was no room for lectures on scientifi c problems, which later turned to be a fatal problem. Interestingly enough, one of the resolutions recommended a general register of the population in each commune, each family being allotted one page where future changes might be recorded.
As to the theory of statistics, a resolution was passed in Florence, in 1867 at the initiative of Quetelet, that there should be created a special section at future congresses, to deal with statistical questions in direct connection with the theory of probabilities. At the following congress, a corre sponding problem was entered, recommending that statistical inves tigations should not only deal with averages, but with the deviations from the mean.
However, the International Statistical Congresses slowly faded away. The rea-sons that led to the end of International Statistical Congresses were realized and taken into account when a few years later the International Statistical Institute (ISI) was established. The fi rst session of the ISI was held in Rome in 1887. Many of the leading statisticians of that time (e.g., Lexis, von Mayr, and Engel) took part in this meeting.