In Thomas Morley's 1597 treatise entitled "A Plaine and Easie Introduction to Practicall Musicke" among various canonic devices, Morely gives an example with instructions on how to compose a palindromic piece for voices. He writes (p.201):
… you may make eight partes in foure (or fewer or more as you list) which may be sung backward and forward, that is, one beginning at the beginning of every part, and another at the ending, and so sing it quite through, and the rules to make it be these:
…because the singer will be in a doubt to which note the dott belongeth. For if he should hold it out with the note which followeth it, it would make an odde number, or then he must hold it in that tune wherein the following note is, making it of that time, as if it followed that note, which would be a great absurdity to let a dott before the note of which it taketh the time: having so made your song, you must set one part at the end of the other of the same kind (as trebble after trebble, base after base &c.) so that the end of the one be joined to the end of the other: soe shall your musicke goe right, forward and backward, as thus for example:
Now Morley's example is just that. He gives the beginning of the solution, and it is clear how it should continue. There is no text, and no source is to be found other than Morley's treatise. But the canon seems only an illustration, as Morley directs the reader to a magnificent example by William Byrd, to be found in the 1575 collection of sacred music by Byrd and Tallis entitled "Cantiones, quae ab argumento sacre vocantur…" a part book. Pictured below is the bass part to this eight-voice canonic palindrome, showing that the part is to be read both forward and in retrograde, with two pages combined into one image (the last line appended):
Next to the initial "D" is the instruction: "Due partes in una recte & retro." [Two parts in one, forward and in retrograde.] Although the text above is inverted, this is not to imply that the retrograde melody is to be inverted as well; it is not.