This is the fifth and final piece of a series on how the adjudication process of the Special Intensive Revision in West Bengal continues in Bengal. Read part one, here, part two, here, and part three, here, and part four, here.The final electoral roll for booth 106 in Samserganj, published on February 28, 2026, lists 1,361 voters at Chandnidaha Primary School in Pratapganj gram panchayat, Murshidabad. Many of the entries carry a stamp across the face of the voters, reading ‘UNDER ADJUDICATION.’The judicial review that followed marked 810 of those 1,361 electors not eligible. That is 427 men and 383 women, and 59.5% of the booth.At this typical booth in West Bengal, about one voter in a hundred was marked not eligible.Across West Bengal the judicial review marked 27,05,027 electors not eligible, spread across 80,719 booths. Half of those booths lost fewer than 10 electors and one in seven lost none at all. One booth in 40 accounts for a quarter of everyone marked not eligible, and one in 12 accounts for half.What this indicates is an un-uniform review that happened intensively in a few places and barely happened elsewhere. Samserganj alone has 36 of the 100 worst-hit booths in the state. Across its 247 booths, it lost 71,983 electors, and its typical booth lost 319 people, where the typical booth in West Bengal lost nine.Counts alone cannot show the picture clearly, because losing 810 from an electoral roll of 1,361 is a different event than losing 810 from a roll of 5,000. Measured against each booth’s own published roll, the worst in West Bengal is part 103 at Dakshin Chakasapur Primary School in Samserganj, where 669 of 1,061 electors – or 63.1% – were marked not eligible. The Chandnidaha 105 booth lost 63%, Chandnidaha 106 lost 59.5%, Panchutola 39 in Baisnabnagar lost 58.5%, and Dakshin Chakasapur 102 lost 58.3%. Seven of the eight highest shares in West Bengal are in Samserganj.Across the state, 49 booths had half or more of their rolls marked ‘not eligible’, 38 of them in Murshidabad and 11 in Malda, with another 255 above 40%. In Samserganj, 30 booths lost half or more and the typical booth lost a third of its roll.Counting by number and by share produces two different lists of booths but the same two districts. Murshidabad holds 63 of the 100 worst by number and 64 by share. For Malda, these numbers are 17 and 18.Three routes off the rollNames came off the roll at three separate stages, and the Commission reported each one separately. First came the clearing of people recorded as ‘absent, shifted or dead’, which took 58,17,851 names off the pre-revision roll of 7,66,37,529 and produced the draft roll. This process required no adjudication. A booth-level officer recorded that a person no longer lived at the address, and the name went from the rolls. Then came objections lodged against individual electors on the draft roll, under Form 7. These removed 5,46,053 names. Then, with 1,88,707 additions, the final list was published at 7,04,59,284. An objection is somebody else asking for a name to be taken off, and it needs no adjudication by a judicial officer either.Last came adjudication. 60,06,000 electors were referred to a judicial officer and 27,16,393 were marked not eligible, leaving 6,77,42,891 entitled to vote on April 9. This is the only stage where an official weighed a file and reached a finding, and it is the one that produced the booth-level concentration described above.Set the three against the minority share of each constituency’s electorate and they do not point the same way. They point in opposite directions, consistently.As the minority share rises, the share of voters marked ‘absent’ or ‘dead’ falls from 8.21% to 5.48%. The share of those removed on somebody else’s objection falls from 0.93% to 0.13, a sevenfold drop. But the share marked ‘not eligible’ by a judicial officer rises from 2.20% to 9.60%. The referral rate behind that last figure rises further still, from 4% of the electorate to 24.48%.At the extremes the gap is wider. The 30 constituencies with the least minority population sent 1.92% of their electors to adjudication and the 30 with the most minority population sent 22.77%, nearly 12 times as many.Sujapur and RaipurSujapur in Malda referred 1,34,521 of its 2,72,097 electors to a judicial officer, which is 49.44% of the constituency. Raipur in Bankura referred 574 of 2,32,460, or one in 405.Sujapur’s electorate is 75.44% minority and Raipur’s is 1.38%. Of the 26,829 Sujapur went on to strike off, 26,752 were Muslim. Raipur struck off 260. Yet, it was Sujapur whose voters could be traced on the old roll more easily, with only one in every 184 unmatched to the 2002 register.Neither of the first two rates is suspicious in itself. Kolkata’s Jorasanko recorded the state’s highest absent-shifted-dead rate at 36.85%, which is what a dense urban seat of short tenancies looks like. What is hard to account for is that these rates had been lowest in precisely those constituencies which were then referred to adjudication in extraordinary numbers.The absent-shifted-dead/deleted or ASDD rate is a measure of genuine churn in an electoral roll. High ASDD means people really did move, die or appear twice, which is what happens in dense urban seats with short tenancies. Low ASDD means a stable, long-settled population. A stable population should throw up fewer documentary problems, so it should generate fewer referrals to adjudication. But we see the opposite in Bengal. An unexplainable large number of voters were sent for adjudication even when these constituencies have the lowest un-mapped names and ASDD numbers.What it looked like on the groundBhatar in Purba Bardhaman shows what happens at the far end of the process. The judicial review marked 17,481 of its electors not eligible. Some 14,413 of them appealed. The Appellate Tribunal has decided 906 of those appeals and refused the elector in 852, a rate of 94% against a state average under nine. That leaves 13,507 people in Bhatar still waiting, in a seat decided by 6,527 votes.Of 17,481 under adjudication, some 14,900 were Muslim electors, or 85.4%, in a constituency whose electorate is 24.7% minority. They are spread across 264 booths in 17 gram panchayats. Bhatar gram panchayat lost an average of 120 a booth. Kurmun-1 lost nine. That is a 13-fold range inside one constituency, under one office, in one revision.What the figures will not supportNone of this establishes that anybody intended the outcome. This is also a mechanism by which it could arise without intent. Adjudication was triggered by discrepancies between an elector’s entry and the documents used to check it. Bengali and Urdu names transliterated into English admit many spellings, and a register compiled across decades by different hands will hold more variants for some names than others. What the series has establishedIn this series, the Election Commission’s own figures have shown that its reply on the appeal tribunals does not reconcile with itself, that seven in 10 of the electors struck off were Muslim in a state where Muslims are under a quarter of the electorate, that the districts which absorbed the most removals have had the fewest corrected, that nine lakh appeals sit unheard in the 49 constituencies where removals already outnumbered the winning margin, and that the two routes requiring no judgement fell hardest in the least Muslim parts of the state while the one route that was entirely based on human judgement fell six times harder in the most Muslim parts, booth by booth.None of that determines an election result, and this series has said so at every stage. What it describes is the situation of 27 lakh people removed from an electoral roll, a remedy that has reached a fraction of them, and a set of questions answerable from records the Election Commission already holds and has not published.For our full coverage on the SIR and the issues it has brought to citizens, click here.