Aphorism 296 is the most honest sentence here; Part Seven isn't patchy, it's bad; read Kaufmann's introduction afterwards or never.

Reader who has arrived here from a recommendation algorithm: this is not the book you want. You want a book of quotations, there is one, it's called something like The Portable Nietzsche, and I'm not sneering, it is a perfectly good way in. This one will hurt you.

Here's my trouble with a six. Sections 1 to 23, the critique of philosophers, the will to truth turned back on itself, the suggestion that every great philosophy is a confession, are among the most exhilarating pages in nineteenth-century prose, and I mean prose, because the sentences persuade some time before the argument does. Then Part Five, the natural history of morals, is patchy. Then Part Seven, on women, is not patchy, it is bad; not bad because it offends me, bad because reasoning that was scalpel-sharp forty pages earlier acquires the quality of a man repeating something he overheard in a spa town. Nietzsche is the last writer who should be excused with "well, his era" — he built an entire method out of refusing that excuse on behalf of everybody else.

The aphorism as a form has a structural problem: it cannot be wrong in public. A paragraph standing alone can never be tested against the paragraph that follows it. La Rochefoucauld understood this and kept his very short, so the reader could see the whole move at once. Nietzsche's run to a page and a half and start to argue, and an argument severed from its objections is a performance of confidence.

And yet. Section 296, at the very end of Part Nine, where he complains that his written thoughts have already lost their colour and gone stale on the page, is the most honest thing in the volume, and it makes the rest retroactively bearable. He knew.

My copy is Kaufmann, which is standard and good, and which carries an introduction I'd advise anyone to read afterwards. Kaufmann is so busy defending his man from the Nazis, a defence that was necessary in 1966 and has since been comprehensively won, that he pre-digests half the difficulties before you have met them. There's a Cambridge edition with Horstmann and Norman doing the translating which I've started using with third-years; the annotation is better and the introduction has the decency to be short.

Six, because the good third is very good, the bad third is genuinely bad, and I refuse to average that into a seven out of politeness. It's an unusual position for me. Normally the parts of a book argue with each other. Here they seem not to have been introduced.

Read the first two parts. Read 296. Then go and read Zarathustra if you must, though I'd rather you didn't.