Picture a scientist as working on part of an enormous crossword puzzle: making an informed guess about some entry, checking and double checking its fit with the clue and already-completed intersecting entries, of those with their clues and yet other entries, weighing the likelihood that some of them might be mistaken, trying new entries in the light of this one, and so on. Much of the crossword is blank, but many entries are already completed, some in almost-indelible ink, some in regular ink, some in pencil, some heavily, some faintly. Some are in English, some in Swahili, some in Flemish, some in Esperanto, etc. In some areas many long entries are firmly inked in, in others few or none. Some entries were completed hundreds of years ago by scientists long dead, some only last week. At some times and places, on pain of firing or worse, only words from the Newspeak dictionary may be used; at others there is pressure to fill in certain entries this way rather than that, or to get going on this completely blank part of the puzzle rather than working on easier, partially filled-in parts-or not to work on certain parts of the puzzle at all. Rival teams squabble over some entries, penciled or even inked in and then rubbed out, perhaps in a dozen languages and a score of times. Other teams cooperate to devise a procedure to churn out all the anagrams of this chapter-long clue or a device to magnify that unreadably tiny one, or call to teams working on other parts of the puzzle to see if they already have something that could be adapted, or to ask how sure they are that it really must be an S here. Someone claims to notice a detail in this or that clue that no one else has seen; others devise tests to check whether he is an especially talented observer or is seeing things, and yet others work on instruments for looking more closely. From time to time accusations are heard of altered clues or blacked-out spaces. Sometimes there are complaints from those working on one part of the puzzle that their view of what’s going on in some other part is blocked. Now and then a long entry, intersecting with numerous others which intersect with numerous others, gets erased by a gang of young turks insisting that the whole of this area of the puzzle must be re-worked, this time, naturally, in Turkish-while others try, letter by letter, to see if most of the original Welsh couldn’t be kept …. I don’t mean to fob you off with a metaphor instead of an argument. But I do mean my word-picture to suggest, what I believe is true, that scientific inquiry is far messier, far less tidy, than the Old Defferentialists imagined; and yet far more constrained by the demands of evidence than the New Cynics dream.