All 34,239 bytes of the tape are accounted for and the BASIC program re-tokenises byte for byte. One thing is still open, and it is a big one.
They sit at 0xA400, 0xAE00 and 0xE323: 1,536 bytes left over behind maps 1 and 2, and a 30-byte tail. They are not bytes nobody touches: line 33 of the program copies each whole scenery, 2,048 bytes, into the work area, and these come along with it.
What is known is mostly what they are not. They have been ruled out as map, colour, music or a copy of any other part of the tape, and as digitised sound. And they have very odd features: only 60 distinct values in 1,566 bytes, a 256-byte template repeated six times, and above all a lead that is as far as anyone has got: the contents seem to be dictated by the memory address they sit at, which on its own explains 85.4% of their bits.
How far that lead goes, and everything measured, is in the dead bytes, with a WANTED poster. To have a go: make misterio extracts and measures them, and they are also loose in datos/misterio.bin.