If a book is predominantly text, the sample is based on a percentage of the number of words (the percentage used is specified in your contract). Where the sample begins can vary depending on how the book has been structured. The words on which the percentage is calculated are counted from the beginning of the book. For flowing books, the beginning is determined by finding the first landmark where the epub:type attribute is "bodymatter". If that attribute is specified, Apple Books starts generating the sample from the publication component in the spine item referenced by the href attribute in the landmark. Then, the sample is generated either from the first item in the landmarks <nav> block, or the percentage of the words, whichever is greater. For example, if the first item in the landmarks <nav> block contains only 2% of the total word count and the percentage specified in the contract is 5%, then the remaining 3% is taken from the next item in the landmarks <nav> block. Any images, movies, or audio included in that percentage will be included in the sample.
If the "bodymatter" attribute is not specified in the landmarks structure, Apple Books looks for the first landmark where the epub:type attribute is not one of the following: "cover", "frontmatter", "toc", "foreword", "introduction", or "dedication" and starts with the spine items immediately following the component referenced in the prior "cover", "frontmatter", "toc", "foreword", "introduction", or "dedication" landmark.
Note: It is recommended that the publication component referenced by the href attribute in the "bodymatter" landmark not be listed close to the last item on the list of spine items. When generating the sample based on percentage of words, Apple Books could potentially include publication components that precede the referenced spine item, which could result in Apple Books including the entire book in the sample.