References

• Paul David Doherty, Infocom Fact Sheet: http://pdd.if-legends.org/infocom/fact-sheet.txt. Section VIII covers the compilations. The Masterpieces subsection is visibly unfinished - it omits Leather Goddesses entirely and has a placeholder in the prose - so prefer disc evidence where the two disagree.
• The part numbers printed on Infocom disks and boxes are Order Codes, documented in Doherty’s Fact Sheet: a Product Code identifying the game plus a System Code identifying the platform, e.g. IM3-CO4 is Suspect for the Commodore Amiga. Section I.1 lists the Product Code of every game; the System Code table gives AP1 Apple II, AP2 Apple Macintosh, AP3 Apple IIgs, AT1 Atari 800/XL/XE, AT2 Atari ST, CO1 C64/128, CO4 Amiga, IB1/IB2 IBM and MS-DOS, TI2 TI-99/4A and two dozen more. After the Activision merger these were replaced by Activision system codes - PD IBM/MS-DOS, TD Apple II, KD Apple IIgs, SD Atari ST, WD Amiga, DD C64/128 - which is why a late disk reads PD-IS3G-04 rather than IS3G-IB2. The four 1988 Solid Gold releases have their own Product Codes ending in G (IZ1G, IC1G, IS3G, IZ0G); Hitchhiker’s Solid Gold is the exception, sharing IS4 with the standard release, which Doherty notes explicitly. The Fact Sheet gives the components rather than an exhaustive title-by-platform table, and not every game shipped on every platform. Its System Code table is also incomplete: a dual-media Leather Goddesses grey box in this catalog carries IC-1-IB3, and the Fact Sheet lists only IB1 and IB2. Treat the table as the best published list rather than a closed set.
• Andrew Plotkin, The Obsessively Complete Infocom Catalog: https://eblong.com/infocom/. Per-game file lists with release and serial numbers, and zcanalyze.py for reading z-code headers. It also hosts a copy of nearly every release, which makes it the reference to check before claiming a build is rare: every build this catalog has read off retail media is listed and downloadable there. Used that way it is also a verifier - all 21 stories extracted here match its copies, 14 byte-identical and 7 differing only in the platform header bytes at 0x01/0x04/0x05.
• The leaked Infocom ZIL sources: https://github.com/historicalsource. Used here for the Zork Zero border analysis.
• Mark Howell’s ztools (txd, infodump, pix2gif): https://ifarchive.org/indexes/if-archive/infocom/tools/ztools/. Fetched and built by scripts/get-ztools.sh.
• ZCut: https://ifarchive.org/if-archive/infocom/tools/zcut11.c. Extracts a story file from an Apple II, Atari 8-bit, Commodore 64/128 or Amstrad CPC disk image, walking each platform’s on-disk layout rather than assuming the story is contiguous. It is the practical way to get a story file off original 8-bit media, and it reports the interpreter letter as a bonus. Two cautions: it needs C99 (enum tsystem is forward-declared through a typedef and wants a small edit to build), and its “Checksum good” is the Z-machine’s 16-bit byte sum, which passes on a permuted image - verify the result independently.
• Greaseweazle: https://github.com/keirf/greaseweazle. Flux-level floppy reader used for every diskette in this catalog; its Flippy Disks wiki page covers reading second surfaces, and issues #460 (Commodore) and #473 (Apple II) document why a flippy’s reverse side reads backwards.
• Michael Sternberg, An Apple II Build Chain for Inform (KansasFest 2017): https://media.unpythonic.net/emergent-files/01765766083/apple_ii_inform2.pdf. Section 2.2 documents how Infocom’s Apple II ZIP/XZIP titles are laid out on disk - 16-sector first disk, boot code and interpreter on tracks 0-3, story from track 5, and the remainder on a second 18-sector image once the story passes 100,864 bytes. It also names the interleaving tools (interl, interlz3, Steve Nickolas’s interlz5) and notes which releases skip interleaving. The only reference found that explains the second surface of a V5 Apple II diskette.
• S. V. Nickolas, interlz5: https://ifarchive.org/if-archive/infocom/tools/interlz5.zip, with Jeff Epler’s Python reimplementation interl5.py alongside it. Converts a Z5 story into Apple II disk images, and its mktrk/mksec are the authoritative description of the 18-sector second surface - one D5 AA AD mark, a 4-and-4 track number, then 18 data fields at 5 + 343*n. Its interl[16] table is the DOS 3.3 interleave applied to side 1’s story area, tracks 4-34 only. Note its two-bit area reads two bytes past each sector, so the last three nibbles of every sector differ from a naive model of the encoder; Epler’s port documents the same inconsistency.
• The leaked Infocom interpreter sources, infocom-zcode-terps: the Apple II XZIP is under apple/xzip/, and rwts.asm holds the reading half of the 18-sector format in a routine commented “READ SECTOR FROM BIG TRACK”. It finds the mark, decodes the track number, then walks sectors forward from it, which is why the 343-nibble stride is exact and no sector can be reached out of order.
• Greaseweazle issue #444: an HD 5.25” drive generates false flux transitions when reading samples as long as 12 microseconds, which is exactly what Apple GCR’s 3-cell gaps (10,020 ns) present. The maintainer’s advice - gw read --densel L - is the single most valuable setting in this catalog: it took three Infocom 18-sector surfaces from almost readable to byte-exact from one capture each.
• IF Archive Infocom media: https://ifarchive.org/indexes/if-archive/infocom/media/. Portable conversions of the Lurking Horror and Sherlock sound files.
• Redump: http://redump.org/. This repo’s pressings of Masterpieces (disc 64685) and The Zork Anthology (disc 82144) hash-match their entries; Redump’s Infocom coverage is otherwise thin - the Lost Treasures volume 1 CDs appear absent entirely.