DfE data tells us that in the period 1995-2006 the number of schools with between 1,000 and 1,499 pupils rose by 35% and the number with between 1,500 and 1,999 by 124%. Following DCSF guidance of this period, local authorities sought to offset this expansion by reducing the capacity of schools unable to offer comparable economies of scale. Correspondingly the number of small secondary schools in England, with fewer than 500 students, fell by 43%. These trends are for the most part unrelated to schools’ educational outcomes: smaller schools are typically closed because they are small, not because they are under-performing; and schools whose performance is average have increased their pupil numbers almost as rapidly as the top performers. A study by Allen and Burgess (2010) comparing the growth rates of English secondary schools always in the top quartile of their local (LA) quality distribution over the past decade with those always in the bottom quartile, and those in the middle half, found that schools in the top quartile grew barely more than the middle half: the median growth rates were 10.3% versus 7.0% respectively. Do the efficiency gains justify the compromise? At one level it’s a ridiculous question, but for complex reasons (explained in depth by Fazackerley et al in their 2010 paper Blocking the Best: Obstacles to new, independent state schools) local authorities have historically preferred the mediocre that they know, to the risks associated with backing new, smaller schools. What’s worse though is that there are real questions whether very large schools do indeed achieve the economies of scale that supposedly justify their expansion. A 2007 PolicyFirst report by Jacob Kestner and Max Haimendorf explored areas where significant diseconomies of scale arise in larger schools. Their analysis highlighted in particular higher communication, coordination, and behaviour management costs and those associated with the accrual of extraneous auxiliary processes and support staff. Of particular concern was the anonymity of many large school environments, which can have an adverse impact on relationships between staff and pupils, and thus on their ability to realise successful educational outcomes. In effect, this means that above a certain size, the per pupil cost increases and to diminishing returns. Kestner and Haimendorf's findings were consistent with the picture that emerged from a systematic review of 31 studies relating to secondary school size (mostly looking at US high schools) by Garrett et al (2004), which found that results and attendance improved with the size of the school only up to a certain point. The problem is that estimates of when dis-economies set in vary widely. The only significant research into performance and school size done on English schools (that of Spielhoferet al (2002) found that the best outcomes were achieved in medium-sized schools with a cohort of approximately 180-200 pupils, which would entail a maximum of 1200 pupils. Working within the constraints of the system, Ketner and Haimendorf argue that when schools adopt the 'characteristics of smallness’ through implementing a house system (each house having say, 150 pupils), for example – which many schools, inspired by The Calouste Gulbenkian Foundation’s ‘Schoolswithin schools’ project, have subsequently done at their recommendation – outcomes may be significantly improved. If you have to expand capacity (and there’s no question of the need to do that), clearly there’s a point in its growth at which the local secondary school stops serving the purpose for which it’s intended, and new schools will be needed. Of course when considering how to allocate basic need capital funding, resources may well go further when targeted at the development of existing facilities. New schools will obviously generally be more expensive to build or (under our present planning system) acquire and convert. But 'playing safe' and channeling funding into expanding already over-sized existing secondaries represents a triumph of short-term financial expediency over what is in the long run going to be best for learners.