For every poor child a grammar school lets in, the school next door takes three: 8.7% on free school meals against 28.3%

Every autumn about a hundred thousand ten-year-olds sit a test that decides which of two kinds of state school they will go to. The Department for Education does not publish who passes. It does publish, for each of the 163 grammar schools that survive from the 1944 system, who is on the roll, what they achieve, how many families put the school first and how many of its places go to children from another council area; councils and consortia publish fragments of the test results. This piece joins all of it: the June 2026 school census, the 2024/25 results and A-levels, the 2026 admissions round, Ofsted's latest judgements, the councils' own test statistics and twenty years of research and policy, for every grammar in England and every one in London and the home counties. The tables for all 163 are attached.¹²³⁶

Three poor children next door for every one inside

The simplest measure of who gets in is the free school meal register. In the 2025/26 census 193,747 pupils were at grammar schools, 5.3% of all state secondary pupils. Of them 8.7% were eligible for free school meals; in the other 3,295 state secondaries, 28.3%. The gap is widest where the grammars are most selective and the counties least poor. The Judd School in Tonbridge has 2.5% of pupils on free school meals and Tonbridge Grammar School 2.6%, in a county whose non-selective schools average 31.2%. Alcester Grammar in Warwickshire 2.2%, King Edward VI Grammar in Chelmsford 2.1%, Sutton Grammar 2.0%, Chesham Grammar and the two Dr Challoner's schools in Buckinghamshire between 2.3% and 2.5%. Twenty-five grammars are below 4%. At the other end the Birmingham King Edward VI foundation, which sets aside a quarter of its places for pupil premium children, has three schools above 27%, and Skegness Grammar in Lincolnshire, at 27.7%, comes closest to its neighbours. The pupil premium measure that the government uses, which counts anyone on free meals in the past six years, tells the same story: 9.1% in grammars, 29.2% elsewhere.²

The government has been trying to move that number for eight years. The 2018 memorandum of understanding with the Grammar School Heads' Association, renewed to 2027, committed the schools to give priority to poorer children, and the Department's own evaluation in 2023 counted the result: the disadvantaged share of grammar pupils rose 'from 7.2% to 7.9%' between 2017/18 and 2021/22, 1,600 pupil premium children among 10,000 extra places, while 136 of the 163 schools now give disadvantaged children some priority and 25 lower the pass mark for them. 'The number of disadvantaged children in grammar schools generally remains low,' the evaluation said. 'The fact that the number … remains low in the face of interventions to increase their numbers shows what a complex issue this is to resolve.' The heads' association reports individual successes, Reading School from 3% to 14%, Queen Mary's in Walsall to 30%, and says 'around 90%' of its members now prioritise pupil premium children. The census shows the national figure a point and a half above where it was.²⁴³⁷²

Nine in ten places leave the borough

A grammar school draws from wherever its test is sat, and the admissions data for this September shows how far that is. Grammars offered 27,350 Year 7 places; families named a grammar as first preference 39,638 times, 1.45 per place; and 7,841 of the places, 28.7%, went to children living in a different local authority. In Kent, which has 32 grammars and its own test, only 6.2% of places leave the county. In London the pattern is reversed, because the capital's grammars are few, famous and reachable by train. Queen Elizabeth's School in Barnet gave 175 of its 192 places to boys from outside the borough, 91.1%; Henrietta Barnett 73.3%; the Tiffin Girls' School in Kingston 70.9%; Latymer in Enfield 70.8%; Beths in Bexley 70.2%; Wilson's and Wallington County in Sutton 67% and 67.6%. Barnet as a borough exports 77.9% of its grammar places, Enfield 70.8%, Kingston 68.8%, Sutton 60.1%, Bexley 50.8%. The exceptions are Redbridge's two schools, which keep three-quarters of their places for borough children, and Bexley Grammar, which keeps 90%. The schools draw their own lines: Newstead Wood in Bromley admits from a nine-mile catchment, and Tiffin in Kingston gives up to 10% of its places for 'aptitude' in sport and music. The pressure follows the same lines: Dartford Grammar had 3.18 first preferences per place, King Edward VI Chelmsford 2.77, Kendrick and Reading School about 2.75, Tiffin 2.46, Queen Elizabeth's 2.25. Nationally 34,785 secondary admission appeals were lodged in 2026 and 5,268, 19.5% of those heard, went the parents' way; the Department does not say how many concerned a grammar.⁶³⁶³⁵⁷

16,446 sat, 8,404 passed, and nobody counts who

Who sits, who passes and who is let in are three different numbers, and only the councils that run the big tests publish any of them. Kent tests more children than anyone: for entry in September 2025, 16,446 sat and 8,404 were 'assessed suitable for grammar', a pass mark of 332 across three papers with no paper below 106. Children at Kent's own mainstream primaries passed at 42.4%; children from outside the county, from Kent's independent schools and other routes, who sat in the same halls, passed at 65.7%. Kent does not separate the private-school candidates from the out-of-county ones and, since 2019, publishes no breakdown by district or school type at all. The adjudicator's 2020 record of the system puts the scale plainly: 'each year the LA tests about 10,000 Kent pupils and about 6,000 pupils from outside the county'. In Bexley, which runs one test for four grammars, 5,866 children sat in 2025 and 2,070 reached the selective standard for 800 places; the top 180 scorers are guaranteed a place and the rest go by distance, and the council notes that 'many' of its top 180 'go to the super-selectives in Bromley or to independent schools'. In Essex the consortium that runs seven grammars has seen candidates rise from 3,572 for 2010 entry to 5,295 for 2025, and publishes the last score admitted at each school each year. Buckinghamshire enters every child at its state primaries automatically and qualifies those scoring 121; the Henrietta Barnett School in Barnet tests everyone and invites only 'the top 300 candidates' to its second round for 120 places.^{9331031111234}

Behind every test is a market. The heads' association's own pupil premium programme, run with the tutoring firm Atom Learning, gives free access to a platform 'normally £575.90 per year' to children on free school meals, and explains why: 'despite encouragement from both primary and secondary schools to parents that the best method of entrance test preparation is mastery of the national curriculum, the most common method of preparing for the 11+ is still intensive coaching and private tuition. This environment ends up providing some children with a comparative advantage, and ultimately widens the educational divide.' Asked last November whether grammar schools should be charging parents for familiarisation tests, the schools minister replied that they 'can charge for optional extras such as familiarisation tests'. The adjudicator has policed the edges: in 2023 the Royal Grammar School in High Wycombe was told its demand that renting families produce a tenancy 'provided by an approved letting agent' was unfair to those in social or temporary housing; in 2024 the thirteen Buckinghamshire grammars were cleared of an objection that their qualifying score was 'fixed and immutable' while being found non-compliant on other points; on 6 October an objection to Churston Ferrers in Torbay was partially upheld over a list of fifteen named feeder primaries.^38263242

Ten times more likely from a prep school

The test is sat at ten and rewards what has happened by ten. The Education Policy Institute's 2016 analysis found that 'by the time the 11 Plus entry exam is taken, 60 per cent of the disadvantaged attainment gap, equivalent to 10 months of learning by this stage, has emerged', and that grammar pupils 'travel, on average, twice as far' as other children to get to school. The Sutton Trust's 2013 study, by a team from the Institute for Fiscal Studies and Cambridge, found that 2.7% of grammar entrants were entitled to free school meals while 12.7% came from outside the state sector, 'largely from independent schools', when just over 6% of ten-year-olds were privately educated; that in selective counties 66% of non-poor children with level 5 in English and maths went on to a grammar against 40% of equally able children on free school meals; and that 'grammar school head teachers believe that children from more affluent, middle class families are coached to pass the entrance exam'. Its 2016 follow-up put it in one line: 'a pupil attending a private prep school is ten times more likely to enter a grammar than a pupil on free school meals'. The same study found disadvantaged Indian children four times, and disadvantaged Chinese children fifteen times, more likely to reach a grammar than disadvantaged white British children. The 2025/26 census shows the result: grammar pupils are 44.4% white British against 60.2% in other state secondaries and 30.9% Asian; at Queen Elizabeth's Barnet 86.8% of pupils are Asian, at Ilford County High 85.6%, at Wilson's 81.8%, at Woodford County 80.9%, at Nonsuch 76.3%. In Kent, by contrast, Dover Grammar School for Boys is 83.6% white British and one in five of its pupils is on free school meals.^1513142

A third of a grade for the few, a penalty for the rest

On raw results the answer looks obvious. In 2024/25 grammar schools posted Attainment 8 scores from the low 60s to the mid 80s depending on the area, against the low 40s to mid 50s for the non-selective schools around them; at least a dozen, among them Wilson's, St Olave's, Henrietta Barnett and King Edward VI Chelmsford, had every pupil pass English and maths at grade 5; and Queen Elizabeth's Barnet topped the A-level table with an average A+ and 84% of entries at AAB or better. The Department published no Progress 8 for 2024/25, because the cohort sat no primary tests in 2020, so the latest value-added measure is 2023/24: Altrincham Grammar School for Boys 1.32, St Olave's and Wilson's 1.27, Queen Elizabeth's 1.22. But Progress 8 compares a child with children who had the same results at eleven, and the research that does that more carefully is less flattering. 'Once prior attainment and pupil background is taken into consideration, we find no overall attainment impact of grammar schools, either positive or negative,' the Education Policy Institute concluded in 2016: a third of a grade more per GCSE for those who get in, 'a small negative effect on those who do not attend grammar schools' in the areas where selection is heaviest, and no net gain nationally because only 5% of pupils are involved. Where grammar places exist for 70% or more of high attainers, the negative effect on the children outside is 'almost 0.2 grades in each of 8 GCSEs' for the poorest, and the gap between free school meal pupils and the rest 'is actually wider in selective areas than in non-selective areas, at around 34.1 per cent compared with 27.8 per cent'. The 2023/24 tables show what that looks like: in Kent the non-selective schools have a Progress 8 of minus 0.35, in Wirral minus 0.34, in Torbay minus 0.42, in Liverpool minus 0.48, in Medway minus 0.16; in London's selective boroughs, where the grammars take fewer of the local high attainers, the comprehensives sit above zero, Kingston 0.65, Barnet 0.59, Redbridge 0.52.^3416152

The institute's 2016 answer to the then government's plan for new grammars was that expanding them 'in areas like Kent, Medway, Thurrock, and much of Buckinghamshire and Lincolnshire would come with a significant risk of reducing attainment in other non-selective schools, particularly for more disadvantaged children', and that of 152 councils only six met its tests for new selection with public support. The prime minister of the day had told the country that 'the attainment gap between rich and poor pupils is reduced to almost zero for children in selective schools'; the institute's reply was that 'highly able pupils achieve just as well in top comprehensives as they do in grammar schools' and that there are 'five times as many high quality non selective schools as there are grammar schools'. The destinations data the Department now publishes shows the sixth-form end of the system: at Queen Elizabeth's 88% of leavers went to a top-third university, at Wilson's 85%, at St Olave's 85%, at Henrietta Barnett 86%; the Department publishes no Oxbridge or Russell Group count by school.^1721165

A ban every government keeps, and an annexe nobody fills

Since the School Standards and Framework Act 1998 no new school may select by ability; the 163 that did so in 1997 were designated grammar schools and may continue, and may be turned comprehensive only by a parental ballot under the Act's own procedure. Expansion is allowed. In 2015 Nicky Morgan approved the Weald of Kent's 'genuine expansion' on to a second site ten miles away in Sevenoaks, noting that '41% of existing pupils' already came from there and that the decision did 'not reflect a change in this Government's position on selective schools'. Ten years on, the adjudicator's determination of 5 August 2026 records what became of the only new grammar school site approved since the 1998 Act: the school 'has the capacity for 2475 children', its published admission number 'had not been reached in 2025/26 or 2026/27', the council confirmed it would not be reached for 2027 either, Year 7 held 260 pupils against a number of 300, and the school is introducing its own 'Weald Grammar Assessment' that 'can be taken in addition to or in place of the Kent Test'. The objector quoted the trustees' minutes: 'constant concerns over the additional costs of having to run a school in two different towns, insufficient applicants to fill available places and proposals to lower the test score to address these problems'; the head had 'evidently unsuccessfully, asked the West Kent Director of Education to discuss lowering the county test 11+ pass mark'. The adjudicator did not uphold the objection to the consultation but found other non-compliances, and noted that 'several other schools in Kent' had set their own tests 'because not enough children were passing the county test to fill the places they have available'.¹⁸¹⁹²⁰

Theresa May's 2016 attempt to lift the ban, with 'up to £50 million a year' for grammar expansion and a promise of 'no return to secondary moderns', died with her majority; the Department's 2018 response confirmed 'the legislation preventing the creation of new selective schools will remain in place' and spent the money instead through a Selective Schools Expansion Fund whose first sixteen winners, from Altrincham to Wolverhampton, 'committed to lowering the mark required to pass the entrance test for pupil premium pupils' in over half of cases. Places offered at Year 7 have grown 9.9% since 2017/18, from 24,354 to 26,776 at the 159 schools present in both years. The present government's position is the same law, stated flatly: 'the government does not plan to allow any new grammar schools to open nor to legislate to remove the right of existing grammar schools to select by ability'. Its Children's Wellbeing and Schools Act 2026 lets councils direct academies to admit a child and lets the adjudicator set admission numbers, and says nothing about selection; its April 2026 answer promised 'new resources to support and encourage schools to use existing provisions within the Code, such as a pupil premium priority'. The Conservative manifesto of 2024 promised only to preserve 'the rights of independent and grammar schools'; Reform UK's contract does not contain the word. In the Lords Lord Watson regretted a Bill with 'nothing to say on the inequality of selection', describing a test 'still inflicted on more than 100,000 10 year-olds in England each year, three quarters of whom are branded failures at that age'; on the Labour benches Rupa Huq called the system 'inegalitarian'. A petition to abolish grammar schools, opened in May, has 52 signatures. In Westminster Hall the live grievance runs the other way: Burnham in Buckinghamshire has a grammar school and, since 2019, no other secondary at all, and 'half the pupils are from Slough'.^{2122236252827394029434130}

What this can and cannot tell you

It can tell you, for every grammar school in England, who is on the roll, how it compares with the schools around it, what its pupils achieved, how many families wanted it and how many of its places went to children from outside the area, from the Department's own files, and what the councils that run the big tests publish about who passes. It cannot tell you who sits the test by family income, because no council publishes it and Kent stopped publishing any breakdown in 2019; nor how many grammar entrants came from private preparatory schools today, because the Department does not record prior school type and the Sutton Trust's figures are from 2013 and 2016; nor whether a particular child would have done better or worse elsewhere, which the research answers only on average. Progress 8 is a year older than the other results because none was published for 2024/25. The comparison groups are all other state secondaries in the same council area, which in a county the size of Kent includes schools far from any grammar. Ofsted's new report cards cover 21 grammars; the rest carry grades under the old framework, some from before an academy conversion. The free school meal measure counts current eligibility, which rose when universal credit widened entitlement, so comparisons with the 2013 and 2016 studies are of direction, not level. Everything else is in the attached appendix, with the method and every table.²³⁸⁹