The standard applied to numbers on this page
This page has been rebuilt after a review found that it ran a provenance audit on figures adverse to the Nizam's state and accepted figures favourable to it, or adverse to the Indian side, on secondary report. That was a real defect and it is worth naming the rule that replaces it.
One rule, stated once and applied in both directions. A figure is used as evidence here only if it can be traced to a document this page has read. A figure that circulates widely but cannot be so traced is named, labelled unlocated, and not relied on — whichever side it helps. On the side adverse to the Nizam's state, that disqualifies the frequently repeated 1941 breakdown of 1,765 officers into 1,268 Muslims and 421 Hindus, the 1911 figures of 70 per cent Muslim police and 55 per cent Muslim army, the reported 1946 Legislative Assembly seat allocation 61, and the published death tolls for the Parkal firing of 2 September 1947 and the Bhairanpalli killings of 27 August 1948 6062. On the side favourable to it, the same rule disqualifies the totals of over Rs 97,000 in temple grants and more than 200,000 acres of inam land, and the 27,000–40,000 range attributed to the Sunderlal report 62. It also requires that Wilfred Cantwell Smith's comparative ratio be discounted on the same terms as the tolls it is used against: Smith's own bases are "it is widely held", "other estimates by responsible observers", and a report he states he had not seen, and the denominator of his ratio — "the number of murders with which previously the Razakars were officially accused" — is not quantified by Smith or anywhere on this page. His sentence is an impression of relative scale, not a measurement, and it cannot be converted into a count.
Where a figure appears on both sides of a casualty or wealth comparison, the same standard is applied to both. The 1947 Commission's jagirdar count and the 1941 census's community proportions are both used to compute over- and under-representation by exactly the same arithmetic, and where that arithmetic runs against the reading this page previously took, it is reported running that way.
One further asymmetry deserves naming rather than quietly correcting. Three of the historiographical positions below — Copland, Datla and Elliott — are reconstructed from other scholars' bibliographies, from a footnote, and from a book review. Kooiman was a fourth until this revision; his 1997 article has now been read directly from the publisher's version of record and his entry is sourced to quotations extracted from it. Those entries are marked as such in place, and the same secondary-report discount applies to them that applies to anything else here.
What the state's own personnel lists actually recorded — and what they did not
The evidentiary spine of this claim is a set of numbers from the Nizam's Civil Lists, and the first thing to establish is what those lists were. They classified officers by place of origin and by salary. They did not classify officers by religion. Karen Leonard, who reconstructed them from the Andhra Pradesh State Archives, is explicit: "Although the Civil Lists continued to classify officials by origin, statistics based upon place of birth no longer reflected social perceptions of group membership" 1.
The 1886 list, the first, showed 476 civil officers: 246 Hyderabadis (52 per cent) drawing 42 per cent of all salary disbursements, and 230 outsiders (48 per cent) drawing 58 per cent — 97 Hindustanis, 66 from Madras, 36 from Bombay, 24 Europeans and 7 from elsewhere. A separate list of the 421 appointments made since October 1884 was starker: 274 of the new appointees, 65 per cent, were Hyderabadis receiving 37 per cent of the new salary bill; 147 outsiders, 35 per cent, received 63 per cent 1. By 1894 the gazetted establishment had grown to 680; non-mulkis had nearly doubled from 230 to 447 while the number of Hyderabadis had fallen to 233 2.
It is at this point that the number which has travelled furthest enters the record, and it enters it from outside the state. Leonard writes: "The best example of the way the Hindu-Muslim issue exacerbated Mulki–non-Mulki tensions comes from a local newspaper's comments on the 1894 Civil List. Noting that only 63 of the 680 were Hindus, and of those only 20 were Mulki Hindus, the article went on to argue that non-Mulki Muslims in the Hyderabad service should not be considered foreigners, for they lent strength to the Nizam's government. Thus religion was combined with the Mulki–non-Mulki conflict in a manner prejudicial to the Hindu majority of the Nizam's population, by reasoning which was to become familiar in Hyderabad" 2. Her footnote identifies the paper as the Deccan Budget of 1 June 1894, held in the clippings collection of the Andhra Pradesh State Archives 3.
How the paper arrived at a religious count from a list that recorded origin is not stated by Leonard and is not known here. Names are the obvious route and this page previously asserted that as established; it is an inference, and it is marked as one. What can be said is that the count has never been confirmed by an official enumeration that anyone has produced, and that it was published in Hyderabad, in Hyderabad's own press, in 1894, so it is not a post-1948 construction. That no contemporary is recorded as contesting the arithmetic is an argument from silence and is labelled as one; the same latitude is not extended to inferences elsewhere on this page, and it should not be extended to this one.
The partisanship of the source is usually treated as a discount. It runs the other way. The Deccan Budget published the count while arguing that non-mulki Muslims in state service should not be regarded as foreigners — that is, while making the case for the outsiders whose dominance the count exposes. Its interest lay in making Hindu exclusion look smaller, not larger. A number published against the publisher's own argument is better evidence than a number published in support of it.
The sharpest form of the figure is one the page previously left unstated. Of the 63 Hindus, 20 were mulkis, and Leonard establishes separately which Hindus the government preferred to deal with: "Almost invariably, the Hindu members appointed to such committees were old city Mulkis from Urdu-speaking communities, showing the government's preference for a certain type of loyal Hindu subject" 16. The mulki Hindu service families were the Kayasths and Khatris of the Mughlai administration, Urdu- and Persian-literate, concentrated in the old city. It follows that the Telugu-, Marathi- and Kannada-speaking Hindus of the districts — the population the claim is actually about, roughly four-fifths of the state — held some fraction of the remaining 43 posts out of 680. The page previously reported 9.3 per cent and moved to complicate it; the figure that matters is an order of magnitude smaller.
The distributional pattern is corroborated elsewhere in the state's own records. Of the seventeen state scholars sent to England between 1897 and 1902, all were Muslims, and critics at the time pointed to the repeated selection of sons of high-ranking non-mulki officials and Muslim nobles 7. The Executive Council formed in 1919 had eight officials among its nine members; mulki nobles held the nominal portfolios while non-mulki officers held Finance, Revenue, Political and Public Works 78. The Classified List of Officers for 1931–32 records that of Osmania University's 71 faculty, 17 gave Hyderabad as their birthplace and 11 were Hindus 9.
The mechanism: origin and language — and why that distinction does less work than it looks
The strongest scholarly objection to reading Hyderabad simply as a state that oppressed Hindus is that the primary axis of exclusion was insider against outsider. Salar Jung I, diwan from 1853 to 1883, needed British-trained administrators and recruited them from north India, Madras and Bombay. Leonard notes that "Most were English-educated, though a few did not know English. There were Parsis and Hindus among them, and some Europeans as well, but most were Muslims and most were from North India", and that after 1857 many Delhi Muslims took jobs in Hyderabad; Aligarh became a major source of recruitment in the 1870s 5.
The insider/outsider structure was not peculiar to Hyderabad, which weakens the reading of the mulki filter as a communal device in disguise and correspondingly strengthens Leonard's reading of it as a sons-of-the-soil politics. Kooiman: 'In Travancore, Baroda and Hyderabad, political struggle was first aimed at a reservation of government jobs for sons of the soil. In this struggle the new, educated administrators imported from other parts of India, whether brahmans or Muslims, played more or less the same part as the white sahib in the British provinces… At first, the sons of the soil were loosely defined in broad categories as Malayalis, Gujaratis and mulkis, but later they were diversified into finer subdivisions of religion and caste' 68. The structural twins are exact — Mysore's 'Mysore for Mysoreans' agitation against Madras Brahmins, Travancore's Malayali Memorial of 1891 — and in all three the sequence runs insider/outsider first and communal subdivision second. The same author draws the opposite conclusion about where Hyderabad ended up; both halves are his and both are reported here.
Government resolutions between 1884 and 1886 then defined the category. "A Mulki was defined as a person who had permanently resided in Hyderabad state for fifteen years or who had continuously served under the government for at least twelve years; he and his lineal male descendants to two generations were legally Mulkis." Leonard's judgement on the effect is unambiguous: "These regulations enabled non-Mulkis and their sons to retain their monopoly on administrative positions" 4. The rules were written by the outsiders they benefited, and they converted the children of outsiders into legal insiders. That matters later in this page, because the same legal category is sometimes used to decide who counts as one of the state's own.
The losers included Muslims, and this is the part of the objection that is straightforwardly right. "The replacement of Persian by Urdu as the language of administration and the courts in the 1880s also worked against the Mulkis, whose specialized knowledge of Persian had been a valued and necessary skill in the old Hyderabad Mughlai administration. The old Hyderabad officials, both Hindu and Muslim, had a decided advantage so long as Persian remained the state's official language" 6. The Mughlai service class — Kayasths, Khatris, old-city Muslim families — was disinherited as a class, not as a creed. No account of Hyderabad that treats the state as a simple engine of Muslim advantage can explain that, and this page does not attempt to.
But the distinction holds at the level of intent and dissolves at the level of effect, and the page's own census section is what dissolves it. That formulation is this page's own argument, constructed from the coextensiveness of the Urdu and Muslim shares, and no specialist has published either it or a rebuttal of it. The nearest published position is Dick Kooiman's, arrived at by a different route and reaching further: he states that he is 'slightly at variance with Leonard's thesis' and that 'with independence drawing near the mulki versus non-mulki issue was gradually subsumed by a larger Hindu-Muslim divide', and that where Leonard reads the mulki/non-mulki educational gap as closing, 'Another conclusion must be that the gap between Muslims and Hindus grew wider' 64. The page's construction should be read as the page's, with Kooiman standing beside it and not behind it. Urdu mother-tongue share and Muslim population share were near-coextensive: 10 and 10 per cent in 1921, 13.4 and 12.8 per cent in 1941 122627. That coextensiveness is treated below as a caution against conflating the two when computing ratios, which it is; but it has a second consequence. Where the language qualification and the confessional boundary enclose very nearly the same people, a state whose administration, courts and only university ran in Urdu had made the linguistic complement of one religion the entry ticket, whatever it intended. For the Telugu-, Marathi- and Kannada-speaking eighty-five per cent, the finding that the axis was Urdu and not creed is a reclassification of the same exclusion, not a refutation of it.
Leonard's own conclusion contains both halves and this page previously quoted only the first. "We have seen that the continued use of the terms Mulki and non-Mulki to designate conflicting groups in Hyderabad has served to over-simplify historical realities. The groups so designated have changed radically over time, in both legal and popular definitions; the most consistent meaning centred on possession of political power." Earlier on the same page she writes: "That the Nizam and other Muslim officials in Hyderabad should perceive Muslims from outside the state as more legitimate participants in government than loyal Hindus and other members of the indigenous population was one of the ironies of these final years" 23. The second is a finding about the regime's own criterion of legitimacy, and it is confessional, in the words of the scholar this page relies on most heavily for the non-confessional reading.
Where the mechanism turns explicitly confessional is datable, and two specialists date it to the same moment. Osmania University, founded in 1918 to modernise the administration, produced instead two rival ideologies among its Urdu-speaking scholars: a "Deccani synthesis" whose slogan was "Long live the Nizam, the Royal Embodiment of Deccani Nationalism", and its opponent, whose slogan was "Long live the Nizam, the Royal Embodiment of Muslim Sovereignty in the Deccan" 1011. The organisation that carried the second was not founded as a communal body. Margrit Pernau: "The Association for the Unity among Muslims, Majlis-e Ittehad-e Musalman, had already been founded in 1927. The first generation consisted mostly of religious scholars who aimed at representing the different schools and sects in Islam and at overcoming the tensions between them. In 1938, Bahadur Yar Jang took over the association and rebuilt it from scratch" 42. Pernau adds that he "was quite ready to alienate the state's Hindu majority with his military parades and grandiose speeches about the Muslims' natural right to rule the country" 43, and that in the constitutional negotiations he intended that "Muslims would keep an absolute majority in every elected body" 42. Leonard's own summary for the same years: "by the late 1930s the ideology of a Muslim state which motivated many high-ranking officials was not compatible with continued popular support by a majority of the state's inhabitants" 22.
Language, schooling and the compounding of disadvantage
Hyderabad's language policy is where structural disadvantage becomes legible for the whole population rather than only for aspirants to office, and here the state's own census is decisive. The 1941 volume records Urdu as the mother tongue of 2,187,005 people and as a subsidiary language of a further 2,238,204 — a combined 27.6 per cent, computed by the census on its base of 16,015,564 speakers of languages indigenous to the state, or 27.1 per cent against the total population of 16,338,534. Telugu was the mother tongue of 7,520,229, Marathi 3,947,089 and Kanarese 1,724,180 27. Roughly seven people in ten could not use the language in which their government, their courts and their university operated.
The capital was demographically unlike the state it governed. Leonard's tabulation of the 1921 returns gives Hyderabad city as 52 per cent Hindu and 43 per cent Muslim, with Urdu the mother tongue of half its residents and Telugu of 39 per cent, against a state that was 85 per cent Hindu and 10 per cent Muslim, with Urdu at 10 per cent, Telugu 48, Marathi 26 and Kannada 12 12. Administration ran on the arithmetic of the capital.
The educational consequences are documented in the state's own enrolment returns, tabulated by S. Rama Char in the Modern Review of August 1939 and combined by Leonard. In 1900 primary enrolment was 25,373 Hindus to 15,324 Muslims, secondary 6,418 to 5,496, college 22 to 12. By 1930 it was primary 180,833 to 99,827, secondary 21,506 to 21,624, college 483 to 667 13. A community that was between a tenth and an eighth of the population held roughly a third of primary places, half of secondary places and a majority of college places. Leonard adds that Muslim literacy "more than doubled from 1881 to 1931, while that for Hindus increased by only 0.4 per cent" — and, in her own footnote, the caveat that this "can be partly explained by the urban concentration of the Muslim population in the state, the expansion of both private and government educational institutions in urban areas, and the higher literacy of non-Mulki immigrants" 1314.
The 1941 census gives literates per thousand as 93 for all communities, 197 for Muslims, 104 for Virashaivas, 63 for "Other Hindus" — the state's single largest category, 9,171,318 people — 11 for Harijans and 8 for Tribes 28. The population-weighted comparison is the one that matters and it runs the claim's way: a minority that was 12.8 per cent of the state was three times as literate as the largest Hindu category and eighteen times as literate as Harijans, and more than ninety per cent of the state's population sat in categories below the Muslim figure.
The same table prints 876 literates per thousand for Brahmanic Hindus, 966 male and 784 female. This page previously built a reframing on that row — that the best-off and the worst-off groups in the state were both Hindu — and the row will not carry it. Applying the volume's own published per-thousand rates to its own published community populations yields about 1,498,000 literates against the volume's stated total of 1,269,004, an overshoot of 18 per cent; the Brahmanic Hindu row is the only line large enough to account for the gap on its own, and forcing it to close the gap gives roughly 246 per thousand rather than 876. Female literacy of 78.4 per cent in 1941 Hyderabad, against 43 per thousand for all communities and 16.8 per cent for women in Travancore in 1931, is not credible on its face. The row is therefore reported as printed and marked as irreconcilable with its own volume, and nothing rests on it. What survives is smaller and still real: a caste and urban gradient runs underneath the communal one, with Brahmanic Hindus (2.2 per cent of the population) and Aryas at 347 per thousand above the Muslim figure and the great mass of the Hindu population far below it. Both gradients are in the table and an account that reports only one of them is incomplete.
On private schooling the record supports less than is usually claimed for it. New rules governing recognition of private schools were issued in 1924, and Leonard attributes them to Arya Samaj schools and gymnasia which "became centres of anti-government propaganda"; but she immediately adds that "many privately-run Hindu schools continued to exist and receive government aid, and the long-established Hyderabad City Arya Samaj continued vigorous religious and social reform activities under the leadership of old city Mulkis" 15. Recognised private institutions fell from 4,053 with 76,654 pupils in 1920 to 929 with 25,865 pupils by 1936, and in Warangal suba 23 of 25 taluqs had neither a middle nor a high school in 1936 46. How much of that fall is closure and how much is reclassification out of the recognised category is not established by the returns alone.
Whatever the mechanism, the outcome is not disputed. Hyderabad's male literacy in 1931 was 8.5 per cent against 17.4 in Mysore, 35.1 in Baroda and 40.8 in Travancore; female literacy was 1.2 per cent against 3.3, 7.9 and 16.8 45. Two literacy series circulate for the state and they do not agree: the 1941 census reports 9.3 per cent literate in 1941 and 5 per cent in 1931 28, while a series compiled from the 1951 census gives 6.8 per cent for 1941 and 4.1 for 1931 44. The discrepancy is a definitional one and readers should expect to meet both numbers.
Worship, patronage, and who actually held the great estates
No evidence of a bar on Hindu worship has been found. That belongs here rather than in the verdict, because no version of the claim under examination alleges one; it is a proposition sometimes attached to this argument by others, and it is answered in passing rather than treated as a finding against the claim. The state's patronage of Hindu institutions is documented from the state's own publications rather than from partisan compilations. Karen Leonard's study of Hyderabad's temples records that a 1943 Information Bureau publication describing prominent houses of worship in the city listed the Kishen Bagh temple with a jagir of Rs 18,000 a year, the Sitaram Bagh temple with a jagir of nearly Rs 50,000 a year, the temple of Jham Singh with a jagir from the third Nizam of Rs 8,408 a year, and the math of Baba Khaim Das with a grant of Rs 11,521 a year; in 1940 the Director of Statistics had published a 239-page list of uruses, melas and jatras in the Dominions 41. Her conclusion is that "Up until the end of the state in 1948, the state administration continued to fund and supervise important Hindu institutions and events, and it publicly proclaimed its role in the face of increasingly communal politics both within and outside the state" 41. The frequently quoted totals of over Rs 97,000 in grants and more than 200,000 acres of inam land have no located archival provenance and are not used here.
The land record was the strongest counterweight this page had, and on inspection it inverts. That correction is the largest single change in this revision and it is worth setting out step by step, because the earlier version stated the opposite four times.
Start with the aggregate, which is not in dispute. The Report of the Royal Commission on Jagir Administration and Reforms, appointed by the Nizam and reporting in May 1947, records that of the Dominions' 82,698 square miles, 40.79 per cent was non-diwani — 9.9 per cent the Nizam's personal sarf-e-khas and 30.9 per cent Paigahs, Samasthans and jagirs — and that of 22,457 villages, 6,535 were jagir villages, 1,961 sarf-e-khas and 13,961 diwani, with roughly a quarter of the state's population living outside direct administration 35. The 1938 Reforms Committee had given a compatible split independently: "the Diwani Ilaqa consists of 58.50 per cent. of the total area and the rest 41.50 per cent. consists of Sarf-i-Khas, Paigahs, Samasthans, Jagirs, and Inams" 32. Taylor Sherman's "43 per cent of land in the state" is closest to the all-non-diwani figure of 40.79 per cent; the Commission's jagir-only figure is 30.9 per cent 5035. These are three different quantities and this page previously printed them side by side as if they were one.
Now the composition. Citing Rai Balmukund's earlier survey, the Commission reports "there are 1,167 Jagirs of which 534 are held by Mulsim [Muslim] Jagirdars, 632 by Hindu Jagirdars" — and in the very next sentence disclaims the figures: "But these figures are entirely out of date and are likely to give a totally incorrect idea of the present situation as regards the number and income of Jagirs in the Dominions." The survey was prepared in 1329 F., "nearly 27 years ago", so around 1919–20 36. A count formally disclaimed by the body that printed it cannot be a headline finding, and the earlier version of this page used it as one.
Even taken at face value it does not say what it was made to say. On the method this page applies to gazetted officers, Osmania faculty, school places and literacy, 534 of 1,167 is 45.8 per cent of jagirs held by a community that was 12.8 per cent of the population — an over-representation of about 3.6 to one. The bare Hindu majority of grants, 54.2 per cent, sits against a Hindu population share above 85 per cent. Read by its own arithmetic, the table is evidence for the claim, not against it.
It is also a count of grants, and the grants are mostly negligible. Of the 1,167, nineteen yielded over a lakh a year, fifteen between fifty thousand and a lakh, and forty-four between twenty-five and fifty thousand; 142 yielded under Rs 500 and 293 under Rs 1,000 36. The Commission's own description of the ordinary jagir is that "the majority of them are holders of single villages without any such power of administration" 37.
The estates that mattered are tabled separately. Paras 61–62 list the seventeen exempted jagirs — the three Paigahs, the estates of the premier nobles, five Samasthans and four other estates — covering 12,218 square miles, 2,934 villages and 1,999,289 people 38. The Commission does not break that table down by community; the following division is this page's own, made by the community of the holding family as the Report names it, and should be read as such. Estates held by Muslim families — the three Paigahs, Salar Jung, Khan Khanan, Fakhr-ul-Mulk, Mehdi Jung and Surya Jung — account for 1,864 of the 2,934 villages (63.5 per cent) and 1,306,102 of the 1,999,289 people (65.3 per cent). Estates held by Hindu families — Kishen Pershad, the Samasthans of Gadwal, Wanparti, Jatprole, Amarchinta and Palavancha, and the estates of Dharam Karan, Raja Shamraj and Kalyani — account for 1,070 villages and 693,187 people. By area the two are close to even, but only because the Palavancha samasthan alone contributes 3,090 square miles of thinly peopled forest against 70 villages; by villages and by population, which is what determines how many people lived under whose collector, roughly two-thirds of those under the great estates lived under Muslim ones. Above them all sat the sarf-e-khas, 8,109 square miles and 1,961 villages, which was not a jagir at all but the ruler's personal estate 35.
So the sentence this page used to run — that most of the great landholders extracting from the peasantry were Hindus — is the inverse of what the cited report records, and it has been struck from the lede, the verdict line and the transmission section. What survives of the older reading is narrower and still true: a very large number of small Hindu jagirdars and inamdars existed; the Telangana rising of 1946–51 was directed at rural intermediaries of both communities; and the people the Nizam's administration served worst were poor cultivators of every religion. The Commission said as much about schooling: "out of a total of over 5,000 schools in the Diwani villages, the Jagir and Samasthan areas have not more than 189" 39. Having noted that a few jagirs and samasthans "provide good facilities for education" but that "these are very rare", the Commission recommended that the Education Department consider whether those particular institutions "should be continued in future as aided schools or taken over entirely by Government" 40; the earlier version of this page broadened that referent to jagir institutions generally, which the Report does not support. The state had fifteen months left.
1938: the ban, the arrests, the procession and the minute, in one place
The government of 1938 did several things in the same year and this page previously separated them by two sections, with the favourable material under a heading reading "where the claim does not hold". They belong together.
In that year the government appointed a committee at the request of the Hindu Civil Liberties Union to investigate the communal riots which had opened the year; it restricted public speeches by particular local leaders on both sides — Bahadur Yar Jung of the Ittehad and Pandit Narinderji of the Arya Samaj; Bahadur Yar Jung was prevented from speaking to a crowd of 15,000 celebrating the Prophet's birthday, while Pandit Narinderji was allowed to lead a Dasara procession of 20,000 sponsored by the Arya Samaj through the city, "and the Nizam himself gave darshan along the way" 21. Leonard's own summary is bounded and the boundary is load-bearing: "the government was not entirely unresponsive, nor was it clearly pro-Muslim, before the Hyderabad satyagraha of 1938" 21.
In the same year the Reforms Committee reported. Chaired by Dewan Bahadur S. Aravamudu Aiyangar with Ghulam Mahmood Qureishi, Professor Qadir Husain Khan, Kashinath Rao Vaidya and Mir Akbar Ali Khan, it had been told by the Nizam that "Its report will be confidential" 34. Its scheme was for functional rather than communal constituencies, and one member's stated attraction to that design was precisely that "under this system no single section will be able to secure a predominant majority for itself" 33. Leonard records the outcome: joint electorates, 42 of 85 legislators elected, and 50 per cent of the seats reserved for Muslims 17.
The Report also prints, in reply to Vaidya's dissent in favour of territorial representation, the minute of Mir Akbar Ali Khan — the fifth member of the committee named above, whom this page previously described only as "a Muslim member". Anonymising him weakened a source that is stronger named. The minute sets out its own conditionality first: "When at first I agreed to the system of representation by interests, I knew that I was throwing the interests of an important minority into serious jeopardy... I was even prepared to face the risks inherent in the functional system without insisting on any reservation of seats or safeguards for the minority community. Now that Mr. Vaidya insists that representation should be territorial, I should like to make it clear that, should that be the case, it would become imperative for me to press the claims for safeguards of the most important minority in the State" 31. It then states the case for minority entrenchment in the state's own voice: "I submit that it will be the height of folly to disregard or forget the history and traditions of a country when its constitution is in the making. A minority which has made an indelible impress of tolerance and justice on the history of the Deccan, and which has laid the foundation of a common nationality and a common culture for its people will never be prepared to sacrifice its influence and power under any constitution merely because it happens to be in the minority." It notes that the late Rai Balmukund had recommended a 50 per cent representation for Muslims, and concludes that "as at the present day, mutual confidence is unfortunately on the wane, it is imperative that equal representation should be granted to the two communities so that the necessary confidence may thereby be restored" 30. The conditionality is worth having in full, and it does not mitigate. The trigger for the demand that a twelve-per-cent minority receive half the legislature was the proposal that seats be allocated by territory — that is, by counting people. Leonard reports that the Report perpetuated "the peculiar political and historical position of the Muslims in Hyderabad" — a phrase which is H. K. Sherwani's characterisation rather than the Report's own wording 17.
A week after the Report was presented, the provisional committee negotiating to form the Hyderabad State Congress was officially notified by the Nizam that such an organisation would be unlawful 18. Accounts date the accompanying Public Safety measure to 6 or 8 September 1938; the notification itself has not been consulted here. The sequence matters and this page previously obscured it. The ban came first, against a locally constituted body formed by residents, before any satyagraha and before any outside volunteers had arrived.
The satyagraha followed in October 1938; after about two months the Congress suspended sponsorship, partly on Gandhi's advice, "because of the increasingly communal character of the satyagraha", and leadership passed to the Arya Samaj, "whose out-of-state members were providing the largest number of volunteer satyagrahis". The government "ultimately arrested some nine thousand people, more than 80 per cent of them non-Mulkis" 19. The earlier version of this page used that 80 per cent to complicate reading the arrests as repression of the state's own Hindu majority. That gloss has been removed for two reasons. The composition of a protest that followed a ban cannot bear on the legitimacy of the ban. And "non-mulki" is a legal category which this page has already shown to have been drafted by the outsiders it benefited, converting the children of immigrants into legal insiders while excluding others 4; a definition built to launder outsiders into insiders cannot then adjudicate who counts as one of the state's own. The figure stands as reported and both facts stand beside it: the mobilisation was substantially organised from British India, and the category doing the sorting was a manufactured one.
Local support did come from within. Hindu shopkeepers observed hartal, and college students sang the forbidden Vande Mataram in their hostels at Osmania and in the district colleges 19. The Deccan Chronicle of 4 December 1938 reported that 115 hostel residents were expelled and that all Osmania students save ten or twelve then observed a hartal 20. The satyagraha was declared withdrawn by the Arya Samaj in July 1939; the ban on the State Congress was not lifted, and "the Reforms of 1938 were not implemented, postponed allegedly because of World War II" 19. A Legislative Assembly is reported to have been constituted in 1946, and accounts in circulation state that its seat allocation gave Muslims more seats than Hindus; the constitutional instrument has not been located and that allocation is not treated here as established 61.
1946 to 1949: the Razakars, the invasion, and numbers nobody has produced
After 1946 the argument moved out of the committee room. Smith, writing in January 1950 after a visit to Hyderabad in spring 1949, recorded the Razakars' "command of funds, arms, vehicles" and "the overt complicity of the minor authorities in its exploits, as well as the tacit approval of those higher up", and that they "terrorized villages, swaggered through the streets of the capital city, looted Hindu shops, raided railway trains, molested women" 48. Sherman adds that through the summer of 1948 they "continued to seek out and eliminate the enemies of the regime" and "targeted not only Hindus, but Muslims whose loyalty was in doubt" 51 — a regime militia widening its target set, not narrowing it. They were also the force deployed against the Telangana peasant rising, which is why the agrarian reading of Hyderabad's collapse and the confessional one describe the same instrument rather than competing with each other.
On scale Smith is cautious in both directions: "Estimates of the numerical strength of the Razakars vary; the figure of 100,000 given by the Indian Government as a minimum, must be considered much too high", and the reports circulating in India "were often exaggerated, and in many instances fabricated, although the plain truth was distasteful enough" 48. Two episodes are commemorated above all others: a police and Razakar firing on a flag-hoisting gathering at Parkal town on 2 September 1947, and the killings at Bhairanpalli on 27 August 1948. Published tolls of about 22 killed and some 150 wounded at Parkal, and of roughly 80 to 200 at Bhairanpalli, circulate in Telangana commemorative writing and press features; no specific publication is named here, no contemporaneous official enumeration underlies them, and under the rule stated at the head of this page they are named and not relied on 6062.
What followed the state's collapse ran the other way, and here this page previously made its worst error of selection. Nehru appointed Pandit Sunderlal and Qazi Abdulghaffar to tour the state. Sherman, who used a version of their report, writes that "the original was suppressed and only scraps of it remain", that they recorded "a widespread anti-Muslim purge" concentrated in Marathwada and Telangana, that "Hindu residents as well as some members of the Army attacked persons and property" — and then, in the two sentences the earlier version of this page hid inside an ellipsis: "Conservative estimates suggest that 50,000 Muslims were killed. Others claim several hundred thousand died." She continues that "Indian troops in some places remained aloof from these activities, in others, they were implicated in them", and that the investigators concluded "In general the attitude of the military officers was good but the soldiers showed bigotry and hatred" 52.
That excision was argument-serving and it is corrected here. What is true, once the full passage and both footnotes are in view, is this. Sherman herself accepts a conservative estimate of 50,000, and her note 46 sources it to "Smith, Hyderabad, p.46", adding that "Smith bases this figure on a report commissioned but never published by the Indian Government" 53. Turning to Smith p.46 closes the loop: "It is widely held that the figure mentioned therein for the number of Muslims massacred is 50,000. Other estimates by responsible observers run as high as 200,000, and by some of the Muslims themselves still higher" 49. The 50,000 therefore rests on what was widely held in 1949 about a report its reporter had not read — which is a weaker foundation than a citation to Sherman implies, and a stronger one than nothing.
Sherman's note 47 attaches not to any 27,000–40,000 range but to "Others claim several hundred thousand died", and disputes K. K. Aziz's use of the Sunderlal report for that claim: "Aziz mentions Sunderlal and Abdulghaffar's report in connection with this figure (p.199-202), but the version of their report cited in this article does not calculate or estimate the total number killed" 54. The earlier version of this page presented that footnote as a demolition of the 27,000–40,000 range. It is not; that range appears nowhere in Sherman. What the footnote establishes is narrower and still useful: the version of the report Sherman worked from contains no total, so any total attributed to the report needs to be traced to a specific page of a specific version.
Extracts from the report were published by Omar Khalidi in 1988 55; the 27,000–40,000 range is attributed in later journalism variously to the report itself and to Perry Anderson, while William Dalrymple has given 200,000, and A. G. Noorani's 50,000 is itself a quotation of Smith 57. None of these is anchored to a located enumeration and none is relied on here.
Smith's one comparative statement is often quoted as an anchor and this page previously treated it as one: "The lowest estimates, even those offered privately by apologists of the military government, come to at least ten times the number of murders with which previously the Razakars were officially accused" 49. Under the rule stated at the head of this page it gets the same discount as the tolls it is set against. Smith's bases are hearsay and a report he had not seen, and the denominator — the official Razakar murder count — is not quantified by Smith and has not been located here. The sentence records a strong contemporary impression that the violence after September 1948 was much the larger of the two. It cannot be turned into a number.
How many Muslims were there? A note on a number this page depends on
Every disproportion computed on this page depends on the Muslim share of the population, and that share is not a single number. The 1941 Census of India, Volume XXI, H.E.H. the Nizam's Dominions gives a total population of 16,338,534 and 2,097,475 Muslims. Imperial Table XIII expresses the communities as proportions per 10,000: Brahmanic Hindus 222, Other Hindus 5,624, Harijans 1,784, Virashaivas 495, Aryas 24 — 81.5 per cent between them — Muslims 1,282, Tribes 415, Christians 135 26. Muslims were therefore 12.8 per cent of the state in 1941.
Leonard's tabulation of the 1921 returns gives the state as 85 per cent Hindu and 10 per cent Muslim 12. The apparent shortfall in that pair is not an error: the 1941 volume counts Tribes as a separate community at 4.15 per cent, and the 1921 census likewise enumerated Animists separately from Hindus. The 1921 volume gives the state total as 12,471,770 with 10,656,453 Hindus, or 85.4 per cent — consistent with Leonard 29.
The movement between the two returns is itself a contested classification. The 1941 volume states that "The corresponding table for 1931 is Table XVI where the distribution of population by Religions was shown. The present classification is according to communities", and records a Muslim variation over 1931 of +36.3 per cent against a Dominions increase of 13.3 per cent 26. A minority growing nearly three times as fast as the state in a single decade, in a census taken across India amid open communal competition over enumeration and abandoned before full tabulation, is a figure to hold loosely, and it destabilises this page's own denominators as much as anyone else's. Wilfred Cantwell Smith, reading the same census in 1950, reported it as "13%, or about 2 million" Muslims and "81%, or about 13 million" Hindus 47.
The practical consequence for this page: the Muslim share should be given as about 10 per cent on the 1921 return and about 12.8 per cent on the 1941 return, and the Urdu mother-tongue share — 10 per cent in 1921, 13.4 per cent in 1941 — should be kept as a separate quantity from it. Conflating them, or averaging them into "a tenth", inflates every ratio computed downstream. That they are separate quantities does not make them separate populations, which is the point argued in the mechanism section above.
How Hyderabad's disproportion compares — the baseline this page previously lacked
A ratio without a baseline invites the reader to supply one, and the one most readers supply is that Hyderabad was aberrant. It was not, in kind. Service monopolies held by small, literate, often imported elites were the princely-state norm, and they ran regardless of the ruler's religion.
In Hindu-ruled Travancore, the state's own recruitment inquiry found the same shape. Dick Kooiman: "In 1934 the Report of the Travancore Public Service Recruitment Committee brought to light that caste Hindus still accounted for seventy-five per cent of government servants, three times their share in the population, whereas ezhavas, as numerous as the nayars, made up only three or four per cent" 65. In Hindu-ruled Mysore, Brahmin preponderance was severe enough that the Maharaja's own committee was told in its terms of reference "that there is at present a large preponderance of the Brahmin Community in the public service", and recommended that within not more than seven years "not less than half of the higher, and two thirds of the lower appointments in each grade of the service" go to non-Brahmins 72. In British India, Muslim members of the Legislative Assembly complained in 1924 that "only some 5 per cent. of the total Government appointments held by Indians are Moslems" 70, and ten years later the Government of India conceded in its Resolution of 4 July 1934 that "these complaints are justified" and reserved 25 per cent of directly recruited vacancies for Muslims 71. The reverse case — a Hindu dynasty over a Muslim majority in Kashmir, with non-Muslims holding the great majority of gazetted posts — is real and is named here, but the Glancy Commission report of 1932 has not been read for this page and reaches it only through a low-quality secondary journal, so under the rule stated at the head of this page it is named and not relied on 73.
What the baseline does and does not do. It does not exculpate. On Kooiman's own figure Hyderabad's Muslims, at roughly a tenth to an eighth of the population, held more than seventy per cent of gazetted appointments — a disproportion of the order of six to one, against Travancore's three to one. Hyderabad is at the severe end of the distribution. What the baseline destroys is a claim of distinctiveness the page never stated but did allow the reader to infer.
There is a second comparison, sharper than the first, which runs the other way. Travancore and Mysore both commissioned official inquiries into their own service disproportions and both were told by their own committees to correct them — Mysore in 1919, Travancore in 1934. Hyderabad's darbar commissioned no equivalent, and the constitutional committee it did appoint moved in the opposite direction, printing a member's demand that the over-represented minority receive half the legislature 30. The comparison relocates Hyderabad inside a common distribution on the numbers, and distinguishes it on what the state did about them.
One further caution, applied to this section as to every other. The page benchmarks Hyderabad against Mysore, Baroda and Travancore on literacy, where Hyderabad looks worst, and did not previously benchmark it on service representation, where the comparison is less flattering to the claim. Selecting the benchmark that favours a conclusion is the same defect as selecting the source that favours it, and it is corrected here rather than left standing.
Three partisan inferences, none official, all pointing the same way
The page's headline count — 63 Hindus among 680 gazetted officers in 1894 — is a newspaper's inference from a list that recorded origin, not religion. It is accepted here on an argument-against-interest rationale, which is legitimate. Two things follow that the page previously did not print.
First, the mirror-image datum exists and must be printed. Kooiman reports that "The Deccan Times on 22 May 1938 took another line in defence of the Muslims" and that on the paper's account "In Hyderabad, Muslims and brahmans were the two chief communities furnishing candidates for government service for generations. Brahmans constituted a negligible minority of two per cent, but held twenty per cent of gazetted offices and forty per cent of others", so that these Brahmans "could not claim to represent the eighty-four per cent Hindus of the state" 66. If accurate, the Hindu share of office was itself concentrated in a caste elite of about two per cent, which would qualify the framing of an excluded 'Hindu majority'. It is discounted heavily and for stated reasons: the Deccan Times was making the darbar's case — a partisan publishing FOR its interest, the exact reverse of the Deccan Budget — Kooiman reports the argument without endorsing its arithmetic, and no official return corroborates it. Discounting it more heavily than the Deccan Budget is legitimate. Printing one and not the other was not, and that is the failure this page was rebuilt to eliminate recurring in a place the last review did not reach.
Second, a third inference exists, from a different decade and a different partisan direction. Kooiman, on the position in the 1930s: "Muslims controlled the government and enjoyed a disproportionate share of gazetted appointments: eight hundred and sixty-four out of a total of one thousand two hundred and twenty-one officers, or more than seventy per cent, leaving the rest to Hindus, Christians and Parsis" 63. His footnote 49 sources the computation to the Mahratta of 10 March 1939 — a Hindu communalist paper from Poona working up 1931 census data. That is a partisan publishing for its interest, and under this page's own rule it earns a heavier discount than the Deccan Budget, not a lighter one. It is therefore not promoted to the lede.
What should carry weight is the convergence rather than any single count. Three separate third-party inferences — Hyderabad's own press in 1894 arguing against its interest, a Poona paper in 1939 arguing for its interest, and the frequently repeated but unlocated 1941 breakdown of 1,765 officers into 1,268 Muslims and 421 Hindus which this page disqualifies for want of provenance — were made from different documents in different decades by parties with opposed interests, and all three put the Hindu share of the higher administration somewhere between a twelfth and a third against a population share above eighty-five per cent. No official enumeration by religion has ever been produced for any year, and the page's finding does not depend on any one of these inferences being right.
The other domain: what the two communities held outside state office
This page computes over- and under-representation in offices, jagirs, school places and literacy to a decimal place, and until now said nothing about how the two communities were distributed across the rest of the economy. That is a chosen denominator and it ran in the page's favour.
Kooiman, immediately after the officer figures: "But on the other hand, Hindus were well represented in the professions and dominated agriculture, trading and banking. The 1931 Census did not even mention trade as an occupation followed by Muslims. For that reason, the darbar tended to consider the Muslim hold on the public services as a kind of justified compensation: since Hindus owned most of the private wealth of the State, 'there are few ways of livelihood left open to the Muslims except Government service...'. And even there, they had to face Hindu competition" 63.
Three qualifications, applied in the directions they actually run. The darbar's inference — that state office was the Muslims' compensating domain — is an interested party's self-exculpation and gets exactly the discount this page gives every other state self-description; Kooiman's own footnote sources it to Copland's reading of the darbar's position, not to a finding. The occupational statement is itself at one remove: Kooiman's note 50 gives Syed Abid Hasan, Whither Hyderabad? (Madras, 1935), p. 75 and Copland (1988), p. 789, so the census claim reaches this page through a 1935 polemic and a monograph neither of which has been read here. And that Hindus 'dominated agriculture' in a state eighty-five per cent Hindu and overwhelmingly agrarian is close to tautological, and does not establish elite economic compensation: on this page's own reading of the 1947 Royal Commission's table, roughly two thirds of the population living under the seventeen great estates lived under estates held by Muslim families 38. The occupational asymmetry is real, it belongs on the page, and it is weaker than the darbar made it.
The ban of 1938: a sequencing dispute between two specialists
This page states, on Leonard, that the ban on the Hyderabad State Congress fell on a locally constituted body before any satyagraha and before outside volunteers arrived. A second specialist reads the same year the other way and the disagreement should be on the page rather than resolved silently.
Kooiman: "When in 1938 the HSC launched satyagraha, the Arya Samaj and Hindu Mahasabha started similar campaigns, posing as champions of the people by their demand for more democracy. The involvement of communal elements in the satyagraha and even in the HSC itself led to an official ban on the HSC", and he prints the government's own stated ground — that it was "confirmed in its conviction that the movement, ostensibly political, [was] in fact a cloak for subversive, communal activities to which the prestige of the name 'Congress' has been deliberately attached for misleading the public" 67.
The government's stated reason is an interested party's and gets no deference — but printing it is required by this page's own standard. The page prints Mir Akbar Ali Khan's minute in full because it is adverse to the state; it cannot then omit the state's own words where they are exculpatory. They are now here.
On the sequence, the two accounts are not evenly matched and this page says which way it reads them. Leonard's order is dated: the notification came a week after the Reforms Committee reported, and the satyagraha followed in October 1938 1819. Kooiman's order is a compressed causal summary carrying no dates, and his own note 54 cites, as the government's further explanation of the ban, "the Chief Minister's address to the Legislative, 23 Aug. 1938" — which places the government explaining the ban some two months before the satyagraha his text says provoked it 67. The page therefore keeps its sequence and records Kooiman's reading as a published dissent rather than adopting it.
What Kooiman does establish, and what the page already records from Leonard, is that communal organisations were substantially involved once the satyagraha was under way: Congress withdrew sponsorship after about two months "because of the increasingly communal character of the satyagraha", and leadership passed to the Arya Samaj 19. The two specialists disagree about causation and about order, not about the presence of the ingredients.
What the page's most-used authority says about the killings after September 1948
Karen Leonard supplies 43.5 per cent of this page's citations, and she holds a published position on the post-invasion death toll which the page did not report. Reviewing Eric Beverley in H-Asia in December 2015 she writes: "While violence took place in some districts after that against Muslims, the casualties were far fewer than rumored, as V. K. Bawa's The Last Nizam (1991), Mohammad Hyder's October Coup (2012), and the finally published Sunderlal Committee Report (in A. G. Noorani, The Destruction of Hyderabad, 2014) make clear" 78.
The interest is disclosed on the same terms as every other: this is a book review, not a researched finding; Leonard offers no figure of her own; and of the three works she names, Mohammad Hyder was the Collector of Osmanabad during the Police Action — a participant, whose account is specific by access and interested by position, exactly as this page treats Wilfred Cantwell Smith and the Deccan Budget. It does not settle anything. It is reported because the page cannot lean on Leonard for its distributional core while withholding from the reader that she publicly dissents from the one figure the page carries at second hand from Sherman.
The second half of the sentence matters more than the first. The Sunderlal–Abdulghaffar report has been in print since 2014 in A. G. Noorani, The Destruction of Hyderabad. This page's open questions previously treated its release as a future event that "would either settle the range or establish that no adequate enumeration was ever made". A printed edition exists and has not been consulted here. That is now the highest-value single acquisition for this page and it is stated as one.