What the alignments of the edition say about the paper of its copies:
how far each copy's paper has stretched along the roll, and how long
the paper of each system is against the reference copy's.
An alignment's scale is one number with three causes. A copy cut for
another paper speed is shorter or longer by the ratio of the speeds,
which the system's copies share; its paper has stretched or shrunk as
it alone has, and so has the reference copy's. In true strains, the
logarithms of the ratios of the lengths, the causes add:
The alignments give the left side to a few parts in a hundred thousand.
The right side has a strain for every copy and a ratio for every
system, so it is found by least squares with each strain taken to lie
about zero by the spread of paper (PAPER_SPREAD, or what the copies
show where there are enough of one system), or about a strain
measured on the copy by that measurement's uncertainty. Where a system
has one copy only its ratio takes the whole scale and the copy's strain
stays at zero, known no better than paper varies; where the copies of
the reference system are several, the axis comes out at their mean
paper and not at the reference copy's.
Copies whose places are not places on the paper, such as a roll
reader's, say nothing about the paper and are left out. Nothing is
said where the edition has no reference copy.
What the alignments of the edition say about the paper of its copies: how far each copy's paper has stretched along the roll, and how long the paper of each system is against the reference copy's.
An alignment's scale is one number with three causes. A copy cut for another paper speed is shorter or longer by the ratio of the speeds, which the system's copies share; its paper has stretched or shrunk as it alone has, and so has the reference copy's. In true strains, the logarithms of the ratios of the lengths, the causes add:
The alignments give the left side to a few parts in a hundred thousand. The right side has a strain for every copy and a ratio for every system, so it is found by least squares with each strain taken to lie about zero by the spread of paper (
PAPER_SPREAD, or what the copies show where there are enough of one system), or about a strain measured on the copy by that measurement's uncertainty. Where a system has one copy only its ratio takes the whole scale and the copy's strain stays at zero, known no better than paper varies; where the copies of the reference system are several, the axis comes out at their mean paper and not at the reference copy's.Copies whose places are not places on the paper, such as a roll reader's, say nothing about the paper and are left out. Nothing is said where the edition has no reference copy.