THE END-EXPERIENCE of Peter Levi's lively extended essay on poems and poets is more like that of having been buttonholed by a talkative and entertaining enthusiast at a pub table rather than that of having attended a formal series of carefully wrought lectures. Levi's knowledge is wide and deep though he carries his learning lightly, and the reader is left with the exhilarating feeling that he has just been led into taking a completely fresh look at every important question concerning the art and craft of poetry. Levi knows that his helter-skelter approach is not only unusual but also tentative and, perhaps, confusing. His often-repeated justification is that he is giving us the sparks struck off in a poet's workshop and not attempting to present a consistent theory worked out fully by analysis: "What I have to say is more haphazard. If this admission needs to be defended, the defence is that my primary concern is poetry, and poets work intuitively, analysis is secondary."

There is a kind of essentialist thesis implied in the title, and this thesis is developed more fully at times, and in distinctly neo-Platonist logic:

What seems extraordinary and most interesting is that so many characteristics of the whole spectrum of poetry from the greatest to the least are determined by such tiny and obvious factors as a repeated noise. And yet it must be so, or we should not be using the one word poetry for such a ...