Compete to crack secrets lost to history—for fun, clout, and (no) profit
The card preserves ordinary letters beside recurrent digits. Its short groups repeat in different sentences: y45 occurs four times as a whole group, g4t twice, and t4 twice. A single assignment has to work across those repetitions, not just make one promising phrase.
The result is an English note in which digits commonly stand for vowels. It recovers more than thirty ordinary word tokens. It does not restore all the handwriting: the lower lines are pale, retraced, or clipped, and the edition keeps those defects visible.
| Written form | Reading | Independent examples |
|---|---|---|
| 1 | a | b1ck → back; d1t2 → date; 1nsw2r2d → answered |
| 2 | e | h2r2 → here; s2nd → send; v2ry → very |
| 3 | i / I | 3n → in; standalone 3 → I |
| 4 | o | g4t → got; p4st → post; l4ng → long |
| 5 | u | y45 → you; c45ntry → country; h5sb1nd → husband |
| 9 | i / I in the recorded contexts | th9s → this; standalone 9 → I |
The upright rendered 1 is a transcription class, not always the numeral one. In a few positions it is the writer’s undotted i or l. Treating every upright as a digit would create false exceptions. The date-like run in line 2 likewise remains numeric; it is not passed through the vowel rule. The key and stopping decision were recorded on 26 August 2026. [1]
The left column is the adjudicated visible rendering. The right column applies the established convention, retaining imperfect source forms. Bracketed explanations distinguish an intended word from what is actually legible.
| Line | Visible rendering | Reading |
|---|---|---|
| M01 | y45 g4t b1ck h2r2 | you got back here |
| M02 | 3n th9s c45ntry 12-1902 | in this country [12-19?2] |
| M03 | 9 g4t y45r p4st c1rd | I got your post card |
| M04 | t4 d1t2. Th4 1nsw2r2d | to date. Tho answered |
| M05 | 9t s4f 3 w3ll s2nd y45 | it so[f/t/∅] I will send you |
| M06 | th1s 4n2. 3 sp4ts 2 | this one. I sp{o}t[s] e [probably “I s’pose”] |
| M07 | y45 1r2 v2ry b9sy / | you are very b[i/9/8]sy [mark] |
| M08 | h5sb1nd c4m2 w1th | husband come with |
| M09 | 9 1m 1t w4ll2 b2n | I am at wolle Ben |
| M10 | 2n y4w h4w l4ng 1r2 | en yow how long are |
| M11 | y45 1g41ng t4 st1y 9 | you a-going to stay I |
| M12 | (on15d.) | ([o/a]n[a/i][u/l]d.) [unresolved] |
Enough survives to say that the writer received a postcard, is sending one in reply, and asks how long the recipient will stay. wolle, en yow, and the final parenthetical are not securely read names or places. “Busy” and “I s’pose” are plausible intentions, not exact letter-for-letter recoveries.
Two independent source readers agreed on all twelve physical lines but initially counted different units. One inventory had 47 groups and 176 expanded slots; the other had 52 groups and 182 slots. A third, source-only adjudication reconciled those inventories against the 3468 × 2298 source image. Its primary rendering retains 52 groups and 182 visible glyph or mark positions, including one slash-like mark of uncertain function. The adjudicator did not use plaintext, a key, or language inference. [3]
The distinction matters: agreement on the outline is not agreement on every letter. The frozen readers matched exactly at 139 of 172 aligned positions before adjudication. The remaining reading problems fall at positions already marked as uncertain in that source-only inventory. Repetition establishes the writing convention; it does not erase those disagreements.
This is a recovered writing convention with a substantial, source-bounded message, not a perfect transcription. HCPortal 1097 is crosswalked to DECODE 3283 by its filename, date, recipient, and places. The bounded prior-solution audit found no public item-level reading in the database, publication, owner-gallery, and indexed-web surfaces it checked. That 26 August search did not cover authenticated DECODE fields. On 28 August, an authenticated review of record 3283 found no attached plaintext, key, transcription, or associated record. The owner’s offline files and unindexed private readings remain outside these checks.[4] The result is a campaign reading, not proof that nobody read the card earlier. [2]
A sharper witness or linked card in the same hand could resolve the residual words. Trying new alphabets against the same faded strokes would not supply that evidence.