Celluloid Skyline: New York in the Movies, by James Sanders

For decades, the world's most famous cityscape was manufactured on a Hollywood lot. Geoff Dyer finds out how movie New York went west ? and finally came home

Walking down Second Avenue once, I noticed a small crowd gathered round the aftermath of an accident. A police car had veered off the street and ploughed through a shop window, wreaking havoc as it did so. I crossed the street to ... Ah, but you know what's coming. That's right: it was a set-up for a film.

This creative collision between two cities – one real, one invented – and, implicitly, a third that results from the synthesis of these two, is the subject of James Sanders' encyclopaedic study. If we imagine his book as a city, then it is huge, sprawling, densely-packed, full of surprising encounters and impressive views. Imagine it as a film, and it is a high-concept project with a big budget and impeccable production values.

More than a decade in the making, Celluloid Skyline is the result of what Sanders modestly terms "an unusual (and relatively extensive) research effort". Much of this effort involved tracking down and trawling through archives of movie stills, almost 300 of which are reproduced. These pictures do not just document or supplement his argument; they also embody it. Impatient readers could get an abbreviated sense of the book's contents simply by studying the pictures and concentrating on the captions. That, if you like, is the filmic way of enjoying it.

On the other hand, the book can be "explored and apprehended" as one would a city: "by wandering through it". Although arranged chronologically, the text can also be dipped into thematically as Sanders focuses his gaze on various aspects of the city's architecture: skyscrapers, tenements, the waterfront, penthouses, townhouses, terraces. Very broadly, Sanders uses each of these building types or features as a screen on which a particular phase or genre of New York's filmic history can be projected.

That history begins in May 1896 when a man named William Heise "initiated a whole new city – through the simple act of recording the existing one". For the next decade, cameramen lugged around their equipment, creating "actualities": little, unedited films that recorded incidents as they occurred.

Almost from the start, these pioneering film-makers sought ways of improving the actual. In 1901 a woman was filmed as warm air from a sidewalk grating blew her skirt up around her legs – an arranged accident more glamorously re-enacted half a century later. It was not long before contrivances like this generated linking stories or plots. Throughout the Twenties many location-shot features were constructed simply by cooking up some kind of scripted incident and tossing it in with the raw ingredients of life in the city.

The introduction of sound made the thriving and thrilling business of shooting in the streets nosily problematic. The solution was to move the city to California, where there was enough space and silence to re-create it in huge studios. When cash-strapped New York writers flocked to Hollywood in the wake of the Wall Street Crash, the imaginative nostalgia created by this migration meant that the invented city became more New York-like than New York. In the horizontal expanse of Los Angeles, this mythic Manhattan appeared more exuberantly vertiginous than ever.

At the same time, the studios went to great pains to make their simulacra as real as possible, importing thousands of reference shots to make every detail convincing. In the Thirties and Forties, the volume of film production was such that, instead of being dismantled at the end of a shoot, the New York set was kept in permanent existence and redeployed in film after film so that the "fake" city took on a familiarity of its own.

Out of this would grow the stunning phenomenon of Hollywood New York: an illusory city in which fantasy and fidelity to the actual were ingeniously entwined. Part of Sanders's skill is to disentwine them, to show how even some of the most fantastic, least socially conscious productions revealed traces of the social reality from which they were offering escape. Crucially, this happened through the sets as much as the scripts. To Sanders, a practising architect and shrewd cultural analyst, walls don't simply have ears – they also have eyes, hearts and an ideological outlook to boot.

Inevitably, the ever-increasing lavishness of Hollywood New York became self-ossifying. The extent to which studio-constructed artifice had grown stale became clear when parts of The Lost Weekend (1945) were filmed on location. As S J Perelman remarked, "the producers sought verisimilitude by bringing Ray Milland to Third Avenue (in the past Third Avenue had always been brought to Ray Milland)."

The move back to the streets of New York was clinched in 1947 when, after much resistance, Mark Hellinger agreed to produce The Naked City on location. Predictably, the shoot was plagued by logistical problems but the voice-over's insistence that "there are eight million stories in the Naked City" persuaded other film-makers of the advantages of trying to tell them on site. This made movies more obviously responsive to what was going on in New York, and in the last third Sanders shows how films like Annie Hall, Taxi Driver or Do the Right Thing did not simply reflect but helped determine the mood of the city. Since these recent films are anchored firmly in the streets of the city, the need for critical mediation is considerably reduced. Sanders's analysis in the last section is less thought-provoking than in early parts.

This book was en route to the printers when the "skyline of lower Manhattan" was "irrevocably altered" last year. Sanders decided to leave it as it was, unaltered. But he was conscious, surely, of the grim irony whereby the enthralling history of New York on film, which began with the little "actualities", reached a stunning, atrocious conclusion with the "actuality" of 11 September.

Geoff Dyer's essays, 'Anglo-English Attitudes', are published by Abacus