The colonial contours of punitive imprisonment in Burundi
The history of the prison in Burundi is linked to colonial penetration, since there were no punitive modalities of confinement in this kingdom for those who broke the laws and social codes before the arrival of the Germans at the end of the 19 th century. In order to grasp the foundations on which the prison system emerged and then developed over a long century, the restitution of its main colonial characteristics is indispensable.
The trace of the chain, or the advent of the prison in German colonial times
In the face of crimes or transgressions, precolonial justice in Burundian society allowed reparations to be arranged by chiefs or bashingantahe, old ‘wise men’ arbitrating local conflicts (Rodegem 1966). This system removed the culprits from their community, and organised ordeals or radical revenge in the most serious cases (Meyer 1984, p. 131). Punitive confinement did not exist and only one exceptional circumstance of ‘detention’ was permitted: the confinement of a prince or chief awaiting the king’s judgment. The transfer of such a prince or chief to the court then required his temporary and supervised accommodation. If the verdict did not lead to his innocence or execution, no sentence implied his confinement: disqualification, confiscation of land or material goods, and the geographical distance of the condemned man were favoured (Simons 1943- 1944, pp. 255-266).
Only during the German occupation (1896-1916) did detention become a punishment in its own right, although it continued to coexist with corporal punishment, deportation, and capital punishment (Chretien 2015, p. 146). In reality, the judgments of the German authorities, which mainly dealt with cases of physical altercation, homicide, or rebellion, did not at the time form part of a desire to ‘modernise’ the justice system. Rather, it was a question of embodying political power over the so-called customary chiefs, whose punitive practices were still being respected in the context of the indirect government of the colonial territories (ibid.).
The detention of individuals thus introduced an innovation in punishment, without however a real penitentiary policy being defined by the German colonial power, neither in the legal field nor in the field of construction (De Wolf2004, p. 334). Thus, while there is evidence of the existence of a prison alongside a military camp in Usumbura (now Bujumbura) in the mid-1900s (Biziyaremye and Kakunze 2011, p. 33), elsewhere it seems that financial issues stifled any prison construction. In Kitega (present-day Gitega), a town founded by the Germans from 1912 onwards with the aim of making it their capital in Urundi, the idea of building cells in the fortified boma building was abandoned after much debate, and it was wooden barracks that housed the prisoners when Belgian troops seized the town in 1916 (Chretien 2015, pp. 69-70,146-147). In the rest of the country, defendants were isolated in military camps or living quarters turned into ad hoc ‘cachets' (literally ‘dungeons’) before they were sent to courts presided over by German officers (Wagner 1999, p. 487).
Ultimately, the Burundian carceral system has kept few traces of the German period, even if the boma, still standing in 2020 and occupied by a police camp, occasionally houses detainees. However, it has retained a vocabulary from that period marked by the beginnings of the application of custodial sentences, and in particular by the physical shackles they implied. Since then, speakers of the Kirundi language have referred to the prison and the prisoner by the term ‘umunyoronf, borrowed from Kiswahili ‘mnyororo’, which means ‘the chain’. Kiswahili was used by the askaris, the African soldiers supporting the German troops in East Africa, who were the first guards assigned to watch over Burundian prisoners, who were systematically chained.1 Upon the arrival of the Belgians, Congolese soldiers from the Force publique (the Belgian colonial army in the Congo) replaced them, perpetuating the use of Kiswahili in the prison environment where the use of the chain only ceased in the mid-1930s. 
The imprint left by Belgian colonisation on the Burundian prison system, on the other hand, is much more significant. Having settled in Burundi after the retreat of the German troops in 1916, the Belgians already had legal texts drawn up for their neighbouring colony of Congo, which they transposed to the territory of Ruanda-Urundi when they took the reins of its administration, adjusting them by means of special orders. This encouraged the extension of a legal arsenal that would increase the number of detainees.
‘Native’ criminalisation and the switch to the prison under the Belgians
Some of the legal instruments which governed the unprecedented development of punitive confinement in Belgian Africa from the 1920s onwards, which have been well studied elsewhere (Dembour 1991, De Wolf 2004, Cornet 2009), need to be clarified in order to understand the situation that prevailed in Burundi with regard to prisons.
First of all, it is necessary to account for the different standards and sentences applied to the ‘indigenes’ (in English ‘natives', i.e., colonised Africans) and other individuals (i.e., Belgians, Westerners, ‘Asians’). The Congolese Penal Code, as applied in Ruanda-Urundi, did deal with “classic” offences such as murder, assault and batter)', and theft, often involving long prison sentences. However, in addition to these, a range of offences charged solely against the ‘indigenes’, introduced from 1918 onwards, also led to short prison sentences, when they were not punishable by fines (Cornet 2009, p. 51). These ‘special offences’ were intended to enforce the new colonial order by punishing breaches of government regulations and administrative rules. They affected the various fields of public order, health and hygiene, trade and agricultural economics, labour, mobility, and taxation (ibid., pp. 57-67). Their implementation led to racial segregation in the distribution of sentences, which has been reinforced by the distinction made in prisons between ‘indigenous’ and ‘non-indigenous’ convicts, to which we shall return later.
Secondly, the courts able to mete out punishments varied, depending on whether a defendant was a native or not and whether the crimes and offences were written law or ‘custom’. In theory, European courts tried non-indigenous people or received appeals from so-called customary courts that dealt exclusively with Africans and ‘applied custom as long as it did not contradict Western justice or morality’ (Cornet 2009, p. 53). But in practice, according to the interpretation of this restriction, colonial courts had greater latitude in their actions than ‘indigenous’ courts and, over time, it appears that, under the pretext of abuse by customary authorities (sometimes very real), colonial justice took precedence over customary law (Gahama 1983, pp. 302-307). A 1926 decree, amended in 1938, and above all the legislative ordinance of 5 October 1943 on the organisation and jurisdiction of indigenous courts, definitively confirmed this superiority (De Wolf 2004, pp. 334-335).
Belgian territorial officials, who were vested with important judicial powers thanks to these texts, used them to ensure the smooth running of their administration, i.e., by punishing many ‘special offences’. Thus, in the image of what the Code de l’lndigenat had encouraged in French colonial Africa, the crimi- nalisation of all kinds of ‘asocial’ (not to say anticolonial) behaviour led to the ‘massification of the use of prison by agents of the colonial administration and, as a result, its trivialisation as a mode of punishment’ (Thioub 1999, p. 288).
Therefore, imprisonment increased. After a decade of Belgian repressive policy, the number of inmates had more than doubled compared to the figure estimated by the Germans in February 1914 (less than 200), with 477 prisoners at the end of 1927, and it had more than tripled some 20 years later, with 715 prisoners at the end of 1936 (Chretien 2015, p. 147, RABRU 1928, 1937). In the aftermath of the Second World War, the number of prisoners remained stable, with 658 prisoners recorded in December 1947,  but in the following decade it jumped to 1,286 prisoners at the end of 1956 (RABRU 1948, 1957). The evolution of the incarceration rate between the beginning and the end of colonisation illustrates this growth in imprisonment, rising from an estimated ratio of 13.5 prisoners per 100,000 inhabitants in 1914 to 63 per 100,000 in 19567
In fact, the number of ‘prison entries’ over a whole year is better able to shed light on prison flows. While 477 men were in prison on 31 December 1927, a total of 1,706 individuals had entered prison that year. In 1936, 715 prisoners were counted on New Year’s Eve, but 3,150 individuals had spent time in prison over the course of the year; on 31 December 1947, the ratio was 358 at that moment versus 1,997 people put in prison over the course of a year (data are unavailable for 1956). Prison numbers were in fact influenced by the length of sentences and the seasonality of incarceration. ‘Special offences’, which accounted for 84.5% of convictions in 1936 and another 55% in 1947 (Cornet 2009, p. 55), were in fact punished by short sentences of ‘penal servitude’ or ‘coercion by body’, so that many individuals spent only a few days, weeks, or months in prison, making them invisible in official statistics. Moreover, the middle of the year was considered to be a ‘peak period’ for incarceration. The dry season (June-August), during the harvesting of the main food crops and coffee, were experienced as a time of euphoria for farmers, whose granaries and pockets were full after the coffee season (Gahama 1983, p. 336, Hatungimana 2005, pp. 378, 413). In ‘this situation of rural jubilation, marked by an increase in food and alcoholic beverages’, the authorities deplored an increase in delinquency, which they said was linked to the agricultural break and beer consumption (Hatungimana 2005, pp. 405-406). The period also coincided with the annual tax collection, which not all peasants were able to pay on time.11 Convictions for public drunkenness, making and consuming fermented beverages, making noise at night, or failing to pay taxes significantly increased the prison quotas.
-  Tying (kuboha) a defendant was not an unknown practice in precolonial Burundi, but it wasdone with shackles (ibohero) of leather or rope (Meyer 1984, p. 131). Kirundi has endorsedthis usage in modern vocabulary, since the words Hmbohero’ or ‘ibohero’ also refer to prisonas a neologism.
-  In 1933-1934, legal texts attenuated the regime of shackling prisoners, but it was Ordinance11/253 of 12 July 1950 that definitively abolished its use in prisons (RABRU 1934, p. 34,1935, p. 31).
-  Rwanda and Burundi were colonised by Belgium as the Territory of Ruanda-Urundi, butgained separate independence in 1962. First placed under a mandate from the League ofNations, then under the supervision of the United Nations (UN), the Territory was administratively united with the colony of Congo by the law of 21 August 1925. Congolese legislationwas made enforceable there by ordinances of the Vice-Governor General of Ruanda-Urundi,abbreviated to VGGRU (Gahama 1983, pp. 44, 415).
-  The Congolese Penal Code of 19 December 1896 was made applicable to Ruanda-Urundi byan ordinance-law of 30 August 1924 and a decree of 10 June 1929; that of 30 January 1940was made applicable by the ordinance of 18 May 1940. It remained in force until independence (De Wolf 2004, p. 335).
-  A complete study of the prison data provided by the Rapports anmtels stir VAdministrationbeige au Ruanda-Urtmdi (RABRU), presented from 1923 to 1961 to the League of Nationsand then to the UN, would be worth undertaking. It is not the subject of this article, forwhich a sampling was carried out of the RABRU 1928, 1937, 1948 and 1957, for which thedata were exhaustive and comparable.
-  The number of prisoners fell by 15% in 1947 compared to the previous year due to morefrequent use of fines (RABRU 1948, p. 190).
-  These rates were calculated on the basis of the German estimate of 1.5 million inhabitants inBurundi in 1910, deemed credible by Thibon (1987, p. 62), and census data provided by theRABRU 1957, establishing the number of inhabitants in 1956 at 2,041,259.
-  The ‘servitude penale principale' (SPP, principal penal servitude) covered the classic prisonsentence, and the ‘servitude penale subsidiaire' (SPS, subsidiary penal servitude) coveredconfinement and labouring to pay off a fine. ‘contrainte par corps' (CPC, coercion by body)consisted of repaying a tax debt through work while in detention.
-  For example, the RABRU 1948 states that failure to provide identification was worth sevendays of SPP, public drunkenness from five days to two months of SPP, and nonpayment oftax from 1 to 14 days of CPC.
-  The phrase was used in a letter from the Governor General of the Congo dated 25 February1952 (ANB Gitega AB 365), who proposed to defer incarceration to avoid prison congestion in the middle of the year.