Social Analysis
Leave a comment

The Elephant in the XUAR: II. Brand new prisons, expanding old prisons, & hundreds of thousands of new inmates

Disciplinary Commission Secretary Yan Bocheng (second from left) during a June 2017 inspection visit to Tumshuq Prison in southern Xinjiang. The prison would start a major facility expansion that year, with an estimated increased inmate capacity in the thousands.

This is the second in a series of three articles highlighting the massive expansion of the prison system in the Xinjiang Uyghur Autonomous Region that has taken place in recent years. The prisons have been running in parallel with the much-covered concentration camps (“vocational training centers”) and possess many of the same traits, interning hundreds of thousands without real due process and engaging in labor exploitation. However, while international action has led to many, if not most, detainees being let out from the camps, those in prisons have been given sentences that often range from 10 to 20 years, and have yet to see any real concessions. The world remains passive on the issue. (Click here to read Part I.)

As the news and stories of long prison sentences have started to come out more frequently over the past two years, it has been natural to perceive this phenomenon as a “recent” development and a “next phase” of the (significantly reduced) camp system. While it is true that some detainees have indeed been sentenced after spending time in camp, the empirical data now available suggest that these are, at best, only a fraction of the current prison inmates. It is probably more accurate to say that the two systems – one extralegal and one (nominally) legal – have been running in parallel since the beginning of the mass incarceration campaign as it started in 2017, with the numbers detained for both in the hundreds of thousands.

The most fundamental official documentation for the scale of the prison internment comes from two independent statistical analyses, done in mid-to-late 2019 by the New York Times and National Public Radio. The New York Times, citing a collation of close to ten official statistical sources, reported approximately 230000 individuals being sentenced in Xinjiang in 2017 and 2018 alone, a five- to ten-fold increase from previous years. One of the sources, an official work report from the Autonomous Region High People’s Court, specifies that in 2018 alone 133198 individuals were sentenced as part of “all out” efforts to protect and ensure “long-term stability”. Meanwhile, National Public Radio, also looking at around five official sources, constructed a graph that showed the number of criminal prosecutions in Xinjiang for 2017 and 2018 totaling to around 350000, also on an order of magnitude greater than in previous years and acutely disproportionate to the rest of the country. With China’s extremely high conviction rate, it is very likely that most of those tried were sentenced.

Given the large number of sources being glued together, the innate fallibility of China’s statistics, and the likelihood of human error – National Public Radio made a major mistake in its initial publication and had to issue an erratum, for example – one should naturally treat these numbers as rough ballpark figures. Working conservatively, we may take the lower (New York Times) figure of 230000 and deduct from it an estimate of what the “normal” amount sentenced per year would be, estimated here as 15000 individuals per year – this is the approximate number sentenced in 2013, as reported by the Xinjiang Yearbook, and corresponds to just before the start of the “People’s War on Terror”. Subtracting the resulting 30000 for the two years then yields a conservative lower estimate of approximately 200000 above-average sentencings for 2017 and 2018 alone. As the mass incarceration campaign since 2017 has predominantly targeted the region’s ethnic minorities, with only a small fraction of those targeted being Han Chinese (traditionally, these have been political dissidents, Falun Gong practitioners, and Christians), it is very likely that almost all of these 200000 above-average sentencings are for those of non-Han ethnicity.

The numbers of documented ethnic-minority victims who were detained in late 2016 or after and who were given prison sentences, by year of sentencing. (Source:; for quality purposes, victims with entries of low/undetermined quality were not included.)

There is very strong reason to believe that the trend did not just end with 2018, either. Using data from the Xinjiang Victims Database, one may take all of the (178) high- and medium-quality victim entries with known sentencing years and look at how they compare. While this sample size is likely too small to allow any strong conclusions about the exact proportions, one does see that the number reported as sentenced in 2019 is comparable to those of 2017 and 2018. If representative, this would then suggest that an extrapolated figure of 300000 may be a more accurate lower estimate. (That very few victims are reported as sentenced in 2020 suggests that either the mass sentencings have ended or that the news has not reached yet.)

To help gauge the societal impact of such a figure, it is useful to consider two additional statistics: the empirical observation that over 80% of the sentenced are men and the official number of adult ethnic-minority men in Xinjiang totaling around 5-6 million. Combining the two with the 300000 lower estimate, it follows that at least 4-5% of the region’s ethnic-minority men are now likely serving prison sentences.

As a visualization, try walking down a busy street and imagine every twentieth man you see sent behind bars for a decade or so. This is, again, based on conservative estimates – the reality may be closer to every tenth.

Another important indicator of the scale of the prison incarcerations can be found in the ratio of those sent to prison to those sent to camp. To this end, one may consider the data from the large cache of local government files obtained by scholar Adrian Zenz, some of which give entire villages of individuals – generally detained in 2017 or 2018 – and explicitly mention each villager’s detention status, usually reported as either camp (送培, 培训中心, 教育转化), prison (判刑), or custody (收押). While the cache itself is still not publicly available, close to 4000 victims from these files have been imported into the Xinjiang Victims Database, thereby making some analysis possible. Of the 1812 victims reported as either being sent to camp or sentenced, 631 (34.82%) belong to the latter, yielding a prison-to-camp ratio of around 1:2. As another 227 are reported as being in police custody, it is likely that the actual ratio is slightly higher, since custody appears to more often be a prelude to prison than to camp (author’s observation).

To see this phenomenon on the microscale, it is helpful to consider a subset of villages from Yarkand County’s Azatbagh Municipality, an area that the available files cover particularly well. With only slight exceptions, the numbers sent to prisons, while generally lower than those sent to camps, are nevertheless comparable and in some places even approach a 50-50 split. Again, as an on-the-ground visualization: imagine that for every three people being detained in these villages in 2017 and 2018, one was given a prison sentence while two were sent to camp.

Numbers of individuals given prison sentences versus those sent to camp for six villages in Azatbagh Municipality, Yarkand County, Kashgar Prefecture. (Source: government spreadsheets obtained by Adrian Zenz and presented in curated form on

Village Name Sentenced Camp Prison-to-Camp Ratio Population (2018)
Azatbagh Village 61 117 1 : 1.92 1043
Chidirtoghraq Village 37 45 1 : 1.22 886
Dong’osteng Village 20 98 1 : 4.90 1494
Lengger Village 57 62 1 : 1.09 1415
Qalpaqdong Village 102 133 1 : 1.30 2555
Qumbolume Village 39 56 1 : 1.44 1032

Virtually all of the victims imported from the cache are from Kashgar and Hotan, and it is fair to ask if such high ratios persist in other – often less politically sensitive – portions of Xinjiang. According to victim records derived from less official sources (i.e., testimonies), it seems that they do, with an analysis of reported victims for the other prefectures across Xinjiang showing that the percentages of detainees sentenced are high there also, including in many of the northern Kazakh areas. While these samples are probably too small to make any precise conclusions at this time, they do seem indicative of the high sentencing rates being a Xinjiang-wide phenomenon.

Numbers of documented victims who have been reported as sentenced in relation to the total numbers of documented victims reported as detained for some period of time since January 2017, by prefecture. (Source:; for quality purposes, victims with entries of low/undetermined quality were not included. Victims were allocated to prefectures by their reported or estimated county of origin.)

Prefecture Sentenced Detained Sentenced %
Aksu 24 59 40.68
Altay 12 27 44.44
Bayingolin 8 26 30.77
Bortala 13 21 61.90
Changji 5 16 31.25
Hami 8 18 44.44
Ili 122 185 65.95
Karamay 3 12 25.00
Kizilsu 18 34 52.94
Tacheng 32 72 44.44
Turpan 5 39 12.82
Urumqi 8 53 15.09

Finally, an arguably weaker source, but one that still helps corroborate the above, is a letter received by Radio Free Asia in September 2018, allegedly written by an Uyghur cadre based in Kashgar (translation available). Though anonymous and difficult to verify, the letter talks about the general situation in the Kashgar prefecture in a manner that is very consistent with the more official documents that came to light later, not only using the same jargon but also referring to a concrete event – a “cautionary film” about the judgment of Uyghur intellectuals – that has since been confirmed. With regard to the scale of the detentions, the author mentions an official document from February 2018 stating that 120000 people had been sentenced in 2017 alone (only slightly higher than the New York Times estimate), while writing that “about 1.5 million people seem to be in the ‘training’ right now, with another 500-600 thousand either sentenced or in detention centers awaiting sentencing”. Again, this is not far from the 1:2 prison-to-camp ratio observed above.

Other reported phenomena, while more qualitative and less direct, contribute to suggesting that the current scale of the prison sentences is such as to overload the traditional prison system.

Among these is the reported transfer of inmates from Xinjiang to inner China, which Radio Free Asia has been able to confirm on at least two occasions through phone calls to the region, in late 2018 with a police officer in Kashgar and again in early 2020 with a police officer in Bayingolin. In an eyewitness account, Memettursun Omer, a cook from Hotan who claims to have been detained in four different detention centers at various times in 2017, recalls how a number of his fellow inmates – particularly, those with heavier sentences – were moved to Henan and other inner China provinces.

Another reported phenomenon is that of deferred sentences. In August 2020, Radio Free Asia was once more able to learn, via phone calls to Korla City’s administrative offices, that 25 people in one of the city’s districts had been given 3-year jail terms to be served in five years, and were currently at home with restrictions. Something similar was also reported by Germany-based scholar Tahir Mutellip with regard to his elderly father, the former Kashgar University professor Mutellip Sidiq Qahiri. After around half a year in detention from 2018 to 2019, the retired professor was released and made to call his son in Germany, threatening to disown him if he didn’t “clean up the shit he made” (by going public about his father’s detention). In February 2020, he was allegedly sentenced to 30 months, with the judgment to be carried out in four years.

In a number of individual documented cases, a person has been sentenced but has continued to remain in the pre-trial detention center for a considerable length of time – in some cases, for years:

  • Asqar Azatbek, a Kazakhstan citizen, was de facto kidnapped by Chinese police while visiting the Korgas International Center of Boundary Cooperation – a visa-free zone on the border between China and Kazakhstan – in late 2017. His court verdict, smuggled out of China, makes no mention of this event but does say that he was sentenced to 20 years in prison in December 2018. In February 2020, sources familiar with the situation reported that he was still being held at the pre-trial detention center in Korgas County.
  • Mahire Nurmuhemmed, a former government administrator from Atush, was allegedly arrested in November 2018 and sentenced to 16 years and 6 months in prison in January 2019, reportedly for sending her son to study Islam in Egypt (and very likely for sending him money while he was there). As of December 2020, her relatives abroad say that she is still being held at a pre-trial detention center in Atush.
  • Oken Mahmet, a Kazakh imam from Altay, was first detained in April 2017, to initially be sentenced 2 months later in June (as testified for by relatives and partially corroborated by a listing on the official Altay court website). In January 2020, his relatives abroad, citing official government notices, reported that he had only been transferred to a prison in Shihezi in November 2019.
  • The “Karakash List”, a government administration document local to Hotan’s Karakash County that was leaked abroad in mid-2019, outlines the fates of hundreds of the county’s detainees. Among the 90 individuals reported as sentenced – generally in 2017 and 2018 – six are reported as being held at a local pre-trial detention center, while an additional two are reported as being held at one of the local camps.

As evidenced by accounts from both inside and outside Xinjiang, pre-trial detention centers in China are generally not good places to be, both because of their poor living conditions and because of their design – unlike camps and prisons, pre-trial detention centers are typically intended for investigation, which in many places has become synonymous with torture and forced confessions (presumably unavoidable when the detainee hasn’t actually committed a crime).

Judging by popular legal opinion (see, for example: here, here, and here), the transfer of a sentenced prisoner from the pre-trial detention center to a formal prison should usually take place no longer than a few weeks or maybe months after the final sentence. As such, the aforementioned cases suggest a sizeable lag in the Xinjiang judicial and incarceration system. One possible explanation is that the appeal process – a right that some do use – has become very slow as the result of a huge influx of cases, and some documented victim cases do suggest this. Bagdat Akin, a Kazakh student in Egypt, was arrested in May 2017 upon his return to China, was sentenced in November 2018, appealed once, had the appeal rejected, and only issued another appeal in July 2019 (having by then spent over two years in detention). Of course, the other explanation for why there exist such large gaps between the sentence date and the actual transfer is the more obvious one – perhaps the prisons just don’t have the space.

That more space is needed has been readily evidenced by the prison facilities’ expansion, enormous in recent years. While still woefully understudied, the expansion of Xinjiang’s prisons, as based on the information currently available, appears to naturally fall into three categories:

  • The construction of brand new prison facilities.
  • The conversion of other (detention) facilities into prisons.
  • The expansion of already existing and established prisons.

Of the 11 formal prison facilities documented by the Xinjiang Victims Database so far – a fairly random, even if small, sample – all but one have expanded as outlined above. 3 of the 11 are brand new – the Kunes Maximum-Security Prison in Ili’s Kunes County (presumably finished in late 2017), the new Fangcaohu Prison compound(s) in the Bingtuan city of Wujiaqu (2016-2017), and the relocated Bingtuan Urumqi Prison on the southwestern outskirts of Urumqi (2016).

The Bingtuan Urumqi Prison, during construction (in 2016) and after.

Of facilities that have undergone conversion to become prisons, only one has been provably identified so far – the new Qarabura Prison in Ghulja City. Originally a drug-rehabilitation center, the now prison has seen a 4- to 5-fold expansion to its premises. A worrisome aspect is the total absence of documentation – it is not possible to identify the facility as a prison from space, and no publicly available documents seem to report it. In fact, the world only knows that this place is now a prison because of an incarceration notice for one of its inmates, Baisultan Yusiphan (sentenced to 15 years), being leaked abroad.

The new Qarabura Prison in northern Ghulja City, formerly a drug-rehabilitation center. Green lines indicate the perimeter of the original center (as of 2015), with the orange showing the current prison compound, the majority of which was expanded in 2017-2019. (May 2020, Google Earth)

This naturally begs the question of how many other such undocumented facilities there are, and additional reports, while less conclusive, suggest that such conversions might not be so rare. While visiting Kashgar in January 2020, Wall Street Journal journalist Eva Dou was told that one facility previously identified as a camp was, in fact, a prison. Meanwhile, in northern Korgas County, an established pre-trial detention center in Twin Channel Village appears to be preparing for an expansion that would double its area, an observation that comes on top of claims that at least five of the victims reported to be held there have all been given prison sentences already. In the absence of additional documentation, one should, at the very least, be on alert for the possibility of this facility too now being converted into a formal prison.

The apparent planned expansion of the pre-trial detention center in Twin Channel Village (Mandarin: 双渠村, Uyghur: Qosh’eriq Kenti) in Korgas County. As seen by comparing the images from May 2018 (left) and March 2020 (right), an area that is approximately equal to that of the original detention center has been cleared and walled off to the south.

Finally, the recent construction of new buildings and/or area expansions of existing prison compounds are easily documented with the help of satellite imagery.

One of the compounds of Shayar Prison, distributed along the Tarim farmlands south of Aksu Prefecture’s Shayar Municipality. The yellow region shows a new second compound, added sometime between 2015 and 2019. (April 2020, Google Earth)

The Xinjiang Women’s Prison in Urumqi. Yellow regions show buildings added since 2016. (July 2020, Google Earth)

Wusu Prison in Wusu City, in northern Xinjiang. Yellow regions show buildings constructed since 2015. Compared to prior years, the farm fields of the southeast compound are no longer green but have been cleared, in what may be preparations for further construction. (June 2020, Google Earth)

The Kashgar Women’s Prison, not far from Kashgar City, was only finished in late 2014, but underwent a major expansion in 2017-2019, as shown by the yellow region. The road that used to run adjacent to the prison’s eastern edge now runs into it. (June 2020, Google Earth)

The Xinjiang No. 3 Prison in Urumqi. Yellow regions show buildings added since 2016. (September 2020, Google Earth)

In some cases, such expansion has been quite dramatic, with a particularly illustrative case being that of the Third Division Prison in the Bingtuan city of Tumshuq. As evidenced by a winning bid notice posted by the China Railway Fourth Engineering Group – a part of the state-owned China Railway Engineering Corporation – the prison underwent a 200-million RMB, 442-day expansion in 2017-2018, adding over 95000 square meters of building space while tripling the size of the compound. About half of that space (over 45000 square meters) was reserved for four identical three-story buildings for prisoner housing and “labor-skills rooms”. Even with conservative estimates – for example, that only a third of the area is used for housing, with the white-paper prescribed 5 square meters per prisoner (and not, say, 2-3, as reported by some ex-camp detainees) – the expansion still implies an increase of at least 3000 in the prison’s inmate capacity.

Translation of the winning bid notice posted by the Seventh Branch of the China Railway Fourth Engineering Group, outlining the details of the Third Division Tumshuq Prison’s expansion (source and original:

Visual expansion of the Third Division Tumshuq Prison since 2017, with the four buildings in red being the new prisoner housing and labor-skills buildings, and the orange line showing the perimeter expansion. Buildings in yellow are new additions that could not be identified with certainty. (Google Earth, April 2019)

Among the 3000+ are several with relatives abroad diligently campaigning for them, with their cases covered in a variety of media outlets and other publications. These include the aforementioned father of Nursiman and Nur’iman Abdureshid – Abdureshid Tohti – sentenced to 16 years and 11 months for “disturbing social order” and “preparing to commit terrorist activities”. Mariye Muhemmed, an owner of an Uyghur restaurant in Boston, USA, has also reported several of her relatives as being held here, including her husband, Sadir Eli – a businessman who has allegedly been given 20 years.

The psychological impact on a person when multiple relatives suddenly go from normal citizens to criminals sentenced to long prison terms is not hard to imagine, and – as discussed in the previous piece – the scenario is not uncommon. However, the phenomenon becomes absolutely devastating when the sentenced is the financial provider for their family. Among the new Tumshuq inmates is also Hesenjan Qari, a textile trader from Atush. Based in Central Asia, where he got married in the 1990s and went on to become a father of six, Hesenjan returned to Xinjiang in 2017 and soon found himself taken to “education” and, another two years later, to Tumshuq to serve a 14-year, 6-month sentence. The charges – “joining a terrorist organization” and “using extremism to undermine the rule of law” – are something his wife, Gulshen, calls a “shameless lie”. Now a single mother of six – five of whom are still underage – she has appealed to both the Kazakhstan and Uzbekistan foreign ministries for help, but only to be told that they can do nothing. Her husband, they wrote, is a Chinese citizen. Or, in other words, China’s “internal affairs”.

Note: This is an updated version of the original article and includes a clarification regarding how to account for the “normal” amount of individuals sentenced independently of the recent mass-incarceration campaigns (specifically, by using the sentencing numbers reported prior to the start of the People’s War on Terror). The original version also overreported the increase in individuals sentenced as being “ten- to twenty-fold” – this has now been fixed to “five- to ten-fold” to more accurately reflect the official numbers.

Leave a Reply