Sequencing errors [081]


Encoding of errors in sequencing, such as scene or page numbering


When an error occurs in the numbering of something reasonably important (e.g. <div>, pages, <lg>, page numbers), the error should be recorded with a <sic> element and its corr= attribute, and the type= attribute of <sic> should be “seq”. E.g., scenes number 1-23, 24, 25, 25, 26, 27, ... would be encoded as

    <div type="scene" n="1"><head>Scene 1</head>

    <div type="scene" n="2"><head>Scene 2</head>

    <!-- ... -->

    <div type="scene" n="23"><head>Scene 23</head>

    <div type="scene" n="24"><head>Scene 24</head>

    <div type="scene" n="25"><head>Scene 25</head>

    <div type="scene" n="26"><head>Scene <sic type="seq" corr="26">25</sic></head>

    <div type="scene" n="27"><head>Scene <sic type="seq" corr="27">26</sic></head>

    <div type="scene" n="28"><head>Scene <sic type="seq" corr="28">27</sic></head>

The same is true even with non-arabic schemes:

    <div type="scene" n="26"><head>Scene <sic type="seq" corr="XXVI">XXV</sic></head>


    <div type="scene" n="26"> <head>Scene the <sic type="seq" corr="Twentysixth">Twentyfifth</sic></head>

Awful Errors in Sequencing

When the sequencing of a system of numbering (e.g. pages, scenes, whatever) is so badly in error that there no longer seems to be any point to using it as a referencing system, we will not record an “idealized” version of the sequence in the n= attribute. To determine how bad the sequencing error is, we use the test of “percentage new error”. This test will count how many new errors occur within the sequence, and calculate the ratio of new errors to total number of items. If one of every ten numbers is a new error (or “10% new error”) we will not record the sequence in the n= attribute, but will only record the actual ink-on-the-page number, using <mw> or whatever element is appropriate.

A new error is an error which is not simply the result of some earlier error. The following examples illustrate the difference between new and old (or “propagated”) errors:

1, 2, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11: this sequence contains 11 items, 1 new error, and 10 old errors (the numbers follow each other correctly, and are only wrong because of the initial repeated number). The ratio of new errors to total items is 1/11, which is less than 10%; hence we would record the idealized sequence on the n= attribute.

1, 2, 2, 3, 8, 6, 7, 8, 9, 10, 11: this sequence contains 11 items, 3 new errors, and 1 old error (the first number 3). The ratio is 3/11, which is greater than 10%, so we would not record the idealized sequence on the n= attribute.

list all entries