Errata to the Second Edition
The publisher prints two hundred of these and binds none of them. They go out folded once, laid inside the front cover, and the first thing a reader does with a book is open it, which is the motion that lets the sheet fall. I have handled eleven copies of the second edition. Three had the slip. One had it as a bookmark at page 140, where nothing on the sheet applies.
What follows is the corrected list, which is to say the list corrected against the list.
p. 12, l. 8. For seperate read separate.
This is the entry everyone reads and it is the only entry that costs nothing. A reader who never sees it is not harmed. I put it first because the compiler put it first, and I think the compiler put it first because it is the kind of error that makes a correction sheet feel like a courtesy and not an alarm.
p. 41, Table II, col. 4. For 2.46 read 1.57.
The table gives the ratio by which the permissible unsupported span improves under reduced gravity. The correct exponent on the gravity term is one quarter. The compiler took a square root. Both numbers are printed in the same book, eight pages apart, because the derivation on p. 33 is right and the table computed from it is wrong.
Nothing on this sheet corrects p. 43, where the argument for the whole chapter rests on 2.46 and is stated as a conclusion. That sentence is not an error of the table. It is a sentence in agreement with a table, and no one checks those. A proofreader compares the printed number against the manuscript number. There is no stage in the making of a book at which a person is paid to ask whether the prose and the arithmetic are describing the same world.
p. 58, eq. 14. For + w²L³/24P² read − w²L³/24P².
The sag correction is subtractive. A steel band hung between two supports lies along a curve, and the distance it reports is the length of the curve, which is longer than the distance between its ends. You are always measuring more ground than there is. The band is honest about its own length and dishonest about yours.
The sign was right in the first edition. It was reset for the second, when the equations were recomposed in a larger face to accommodate the new plates, and somewhere between the compositor’s stick and the proof the mark inverted. Two county boundaries were run with the second edition open on the tripod case. In both the error is under an inch in a chain and neither has been re-run, because re-running a boundary requires someone to want the boundary moved, and the person who wants it moved is by definition the person losing.
p. 59. No correction. The paragraph on normal tension stands as printed.
I record it here because it is the finest thing in the book and a reader who comes to this sheet looking for damage will skip it.
The tape hangs in a curve and reads long. Pull it harder and the curve flattens, and the reading comes down. Pull it harder still and the steel itself stretches, and the reading goes long again for an entirely different reason. Between those two, at one particular pull, the shortening you take out of the sag is exactly the lengthening you put into the metal, and the tape reads true. It is not straight. It has not been corrected. It is wrong in two directions at once and the wrongness has been tuned until it cancels, and the compiler’s word for this condition is normal.
Every instrument I have ever trusted has been in that state and has not said so.
p. 88, dedication. For 1934 read 1935.
The compiler of the second edition dedicated it to the compiler of the first, who was his teacher, and got the year of his birth wrong by one. The family wrote in. The letter is in the publisher’s file and is four lines long and does not complain. It says the year, gives the parish, and offers to supply the register entry if it would help.
This is the only entry on the sheet that corrects the world instead of the mathematics, and it is the only one where a correction changes nothing. The man was born when he was born. What the errata repairs is the book’s claim about him, and the book will go on saying 1934 in every copy that has lost its slip, which is most of them.
pp. 101–118. No entries.
The chapter on cable laid directly on the ground gives no correction for slip at the points of contact. The reader is told that friction is sufficient and referred to a footnote which refers to a paper which states, in its final paragraph, that the question was not addressed.
An errata sheet has no grammar for this. The form is for X read Y, and the form requires an X. You cannot write for nothing read something; the compositor would have no line to set. So the largest error in the second edition is not on this sheet, and could not be, and the machinery that produced this sheet is constitutionally unable to notice it. The corrections that get made are the corrections that fit the correcting instrument. That sentence is true of the tape on p. 58 as well, and I did not notice the parallel until I had set both entries in the same afternoon.
I have spent four months on the omission and I can tell you what the answer depends on and not what it is. This is why the entry above says no entries instead of forthcoming.
The sheet itself.
Two hundred were printed. The edition ran to eleven hundred. The compiler requested that the corrections be carried into a third printing and there was no third printing, so the second edition is the edition, and it is wrong on p. 41 and p. 58 in every copy, and right on p. 59 where nobody looks.
The error is bound in and the correction is loose. This is not an oversight of the trade. Binding costs money at the signature, and a correction discovered after the sheets are folded arrives too late to be sewn into anything. So the truer half of the book is the half that falls out, and the half that survives is the half that was printed before anyone had checked it.
If you have this sheet, copy it onto the flyleaf in pencil. Pencil is what surveyors use in the field, for the reason that everything written in the field is provisional and the ones who go out in weather learned this before the ones who set type.
Corrections to this sheet may be sent to the publisher. There has never been one. I do not take that as evidence.
