limited to roughly 30 other languages. It is possible that some stateless persons speak languages for which no assistance is available. Individuals with insufficient language skills may have difficulty completing the survey. Without more reliable and detailed data, the scale and direction of biases in CMS’s estimates cannot be determined. The data from US refugee and immigration programs feeds into an extensive, interlocking set of federal databases and information systems (Meissner et al. 2013, 66-72). DHS has initiated an ambitious data integration project that seeks, in part, to ensure that immigration records can be linked across the various DHS data systems, and made accessible to DHS stakeholders. Stateless persons may come to the attention of federal officials that administer US refugee and immigration programs through applying for an immigration benefit, admission to the United States, refugee resettlement, requesting asylum, immigration enforcement, placement in removal proceedings, detention, and other ways. At each of these points of contact, non-citizens must provide their “nationality,” which may be listed as “stateless,” “unknown,” or possibly “no nationality.” Ultimately, CMS researchers opted to build their provisional estimates around ACS data given its reach and detail, and because of the limitations of federal immigration and refugee data, starting with the fact that not all stateless persons come to the attention of immigration officials. After significant engagement with DHS on this study and their data needs, CMS formally requested from DHS detailed information on admissions of refugees, non-immigrants, and immigrants (LPRs), as well as anyone listed as stateless or of unknown nationality in the same datasets. CMS also requested detailed information on non-citizens—particularly those listed as stateless or of unknown nationality—who received removal orders, and either spent more than 90 days in detention following their order or who were released from custody. As per the latter request, DHS is statutorily required to remove a non-citizen within 90 days following an administratively final order of removal.62 In Zadvydas v. Davis,63 the US Supreme Court found six months to be a presumptively reasonable time-period to effect removal. After six months, however, it held that the government bears the burden to demonstrate that the detainee is likely to be removed in the “reasonably foreseeable future.” If it cannot meet this burden, it must release him or her. In Clark v Martinez,64 the Supreme Court extended this decision to inadmissible non-citizens; i.e., those encountered by immigration officials before they “enter” the nation as a matter of law. Following Zadvydas and Clark, the number of non-citizens, including stateless persons, detained for very long periods following receipt of a removal order, declined significantly (Kerwin, Alulema, and Tu 2015, 342). That said, data on non-citizens who cannot be removed still remains a likely source of stateless persons. A study by the DHS Office of Inspector General (OIG) of all immigrant detainees on December 13, 2017, for example, found that 3,053 had not been removed within 90-days of their removal order (DHS-OIG 2019). This figure exceeded the numbers reported in earlier studies that examined all US immigrant detainees on particular nights in 2009 and 2012 (Kerwin and Lin 2009, 18; Kerwin, Alulema and Tu 2015, 3 43). Moreover, the OIG report attributed delays in 62 INA § 241(a)(1)(A). 63 533 US 678 (2001). 64 543 US 371 (2005). 60