“But it’s always crooked,” said Flagg, “and I have a great fancy for the line it makes.”

She straightened it on Clemens himself, but it immediately became crooked again.  Clemens said: 

“If you were to make that necktie straight people would say; ’Good portrait, but there is something the matter with it.  I don’t know where it is.’”

The tie was left unchanged.

CLXXIV

THE MACHINE

The reader may have realized that by the beginning of 1891 Mark Twain’s finances were in a critical condition.  The publishing business had managed to weather along.  It was still profitable, and could have been made much more so if the capital necessary to its growth had not been continuously and relentlessly absorbed by that gigantic vampire of inventions—­that remorseless Frankenstein monster—­the machine.

The beginning of this vast tragedy (for it was no less than that) dated as far back as 1880, when Clemens one day had taken a minor and purely speculative interest in patent rights, which was to do away with setting type by hand.  In some memoranda which he made more than ten years later, when the catastrophe was still a little longer postponed, he gave some account of the matter.

    This episode has now spread itself over more than one-fifth of my
    life, a considerable stretch of time, as I am now 55 years old.

Ten or eleven years ago Dwight Buell, a jeweler, called at our house and was shown up to the billiard-room-which was my study; and the game got more study than the other sciences.  He wanted me to take some stock in a type-setting machine.  He said it was at the Colt’s Arms factory, and was about finished.  I took $2,000 of the stock.  I was always taking little chances like that, and almost always losing by it, too.  Some time afterward I was invited to go down to the factory and see the machine.  I went, promising myself nothing, for I knew all about type-setting by practical experience, and held the settled and solidified opinion that a successful type-setting machine was an impossibility, for the reason that a machine cannot be made to think, and the thing that sets movable type must think or retire defeated.  So, the performance I witnessed did most thoroughly amaze me.  Here was a machine that was really setting type, and doing it with swiftness and accuracy, too.  Moreover, it was distributing its case at the same time.  The distribution was automatic; the machine fed itself from a galley of dead matter and without human help or suggestion, for it began its work of its own accord when the type channels needed filling, and stopped of its own accord when they were full enough.  The machine was almost a complete compositor; it lacked but one feature—­it did not “justify” the lines.  This was done by the operator’s assistant.

I saw the operator set at the rate of