diff --git a/crates/gc-extract/tests/disc.rs b/crates/gc-extract/tests/disc.rs
index 32681e5..51a8f74 100644
--- a/crates/gc-extract/tests/disc.rs
+++ b/crates/gc-extract/tests/disc.rs
@@ -6539,3 +6539,85 @@ fn tie_class_header_tail_words() {
     );
     eprintln!("0x3c histogram: {w3c:?}");
 }
+
+/// `0x1c` and `0x34` of the tie class header are base-relative offsets, and they
+/// land past the class's own tie packets.
+#[test]
+fn tie_class_tail_offsets_point_past_the_tie_packets() {
+    require_disc!();
+    let data = std::env::var_os("GC_EXTRACTED")
+        .map(PathBuf::from)
+        .unwrap_or_else(|| {
+            let home = std::env::var("HOME").unwrap_or_else(|_| ".".into());
+            PathBuf::from(home).join("Documents/gc-extracted")
+        });
+    if !data.exists() {
+        eprintln!("skipped: unpack the disc first");
+        return;
+    }
+    use gc_formats::level_core::{ClassEntry, LevelCoreHeader};
+    use gc_formats::tie::TieClassHeader;
+
+    let (mut classes, mut in_range, mut past_packets, mut both_nonzero) =
+        (0usize, 0usize, 0usize, 0usize);
+    for lv in 0..27usize {
+        let (Ok(idx), Ok(core)) = (
+            std::fs::read(data.join(format!("level{lv}/core_index.bin"))),
+            std::fs::read(data.join(format!("level{lv}/core_data.bin"))),
+        ) else {
+            continue;
+        };
+        let t = LevelCoreHeader::parse(&idx).expect("hdr").tie_classes;
+        for c in 0..t.count as usize {
+            let o = t.offset as usize + c * 0x20;
+            let Ok(e) = ClassEntry::parse(&idx[o..o + 0x20]) else {
+                continue;
+            };
+            let base = e.offset_in_asset_wad as usize;
+            if base == 0 || base + 0x48 > core.len() {
+                continue;
+            }
+            let Ok(hdr) = TieClassHeader::parse(&core[base..base + 0x48]) else {
+                continue;
+            };
+            classes += 1;
+            let word = |k: usize| -> u32 {
+                let at = base + 0x1c + k * 4;
+                u32::from_le_bytes(core[at..at + 4].try_into().unwrap())
+            };
+            let (o1c, o34) = (word(0) as usize, word(6) as usize);
+            if usize::from(o1c) + 32 <= core.len() && o34 + 32 <= core.len() {
+                in_range += 1;
+            }
+            if word(0) != 0 && word(6) != 0 {
+                both_nonzero += 1;
+            }
+            // Past the class's own tie packet blocks.
+            let last_packet = hdr
+                .packets
+                .iter()
+                .filter(|p| **p > 0)
+                .max()
+                .copied()
+                .unwrap_or(0);
+            if o1c as i32 > last_packet && o34 as i32 > last_packet {
+                past_packets += 1;
+            }
+        }
+    }
+
+    assert!(classes > 1_800, "only {classes} tie classes read");
+    eprintln!("0x1c/0x34 in range {in_range}/{classes}, past own packets {past_packets}/{classes}");
+    // Both are relative to the class base, not absolute: base + offset lands in
+    // core_data, and absolute offsets would not.
+    assert!(
+        in_range * 100 >= classes * 99,
+        "only {in_range}/{classes} offsets in range"
+    );
+    assert_eq!(both_nonzero, classes, "a class has a zero offset");
+    // They are never inside the class's own tie data.
+    assert_eq!(
+        past_packets, classes,
+        "{classes} offsets fall inside the tie packets"
+    );
+}
diff --git a/crates/gc-formats/src/tie.rs b/crates/gc-formats/src/tie.rs
index 5282b30..a4032a7 100644
--- a/crates/gc-formats/src/tie.rs
+++ b/crates/gc-formats/src/tie.rs
@@ -111,8 +111,14 @@ impl TieClassHeader {
         //               bfc_distance is flat across every bucket.
         //   0x40        a float -- 0.00208 to 10.84, no negatives. The bounding
         //               sphere radius or the class scale; not distinguished.
-        // 0x1c, 0x34 look like offsets (0x1e0..0x3cd60, 0x190..0x31e20) and
-        // 0x24, 0x28, 0x2c, 0x38 are high-entropy. Unidentified.
+        // 0x1c and 0x34 are base-relative offsets: base + offset lands inside
+        // core_data on 1,864 of 1,864 classes, and always *past* the class's own
+        // tie packet blocks, so they address per-class data stored elsewhere.
+        // The bytes there are structured -- repeating small-integer records,
+        // occasional `ff 00 ff ff` -- but are not VIF or GIF DMA tags and are
+        // not identified. See tie_class_tail_offsets_point_past_the_tie_packets.
+        // 0x24, 0x28, 0x2c, 0x38 are high-entropy and unidentified; 0x24 is
+        // demonstrably not an offset (17 of 1,864 land in range).
         b.skip(0x44 - 0x1c, "tie class fields not decoded yet")?;
         let o_class = b.i32("tie class number")?;
         Ok(TieClassHeader {
diff --git a/docs/plan/level-assets.md b/docs/plan/level-assets.md
index 97b8408..95ab69e 100644
--- a/docs/plan/level-assets.md
+++ b/docs/plan/level-assets.md
@@ -1849,3 +1849,41 @@ vertices, scissor. `bfc_distance` is a per-record quantity relating to the shade
 count whose meaning is not established, and the "quadwords of ad-gif data" reading
 of it is wrong. The field doc said so in terms before this measurement; it does
 not now.
+
+
+## Where the AD-GIF is not: class-header offsets 0x1c and 0x34
+
+Two words in the class header's undecoded tail are offsets, and they are the
+best remaining lead on the AD-GIF.
+
+`base + 0x1c` and `base + 0x34` land inside `core_data` on **1,864 of 1,864
+classes**, so both are relative to the class base rather than absolute, and both
+are non-zero on every class. And both land **past** that class's own tie packet
+blocks on every class -- they address per-class data stored somewhere else in
+core_data, not the tie geometry.
+
+What is there is structured, not random. Class 0 at 0x1c:
+
+    54 5d 5f 0c  44 44 44 01  03 11 19 00  1c 00 60 0a  30 08 60 0a
+    ff 00 ff ff  20 0c 50 02  44 38 2b 50
+
+and class 3 at the same field:
+
+    20 05 c5 00  00 80 01 00  26 05 cb 00  21 88 01 00  2c 05 c0 00
+    00 80 01 00  32 05 c0 00  00 80 01 00
+
+Class 3 is a fixed-stride table with `80 01 00` recurring. Class 5's 0x34 field
+has `38 38 38` and `ff 00 ff ff`. Class 2's 0x1c field is a monotone byte table:
+`26 27 0e 0f 10 11 12 13 24 25 15 1e 17 1f 14 15 16 17 20 21 22 23`.
+
+These read like compiled shader programs with small-integer parameters, or as
+lookup tables. **They are not identified**, and the resemblance to anything
+specific is exactly the kind of finding rule 4 exists to refuse -- none of these
+three fragments has been shown to beat a control, and they were picked because
+they looked like something.
+
+What is established is structural: two base-relative offsets, always in range,
+always past the tie packets, pointing at repeated structured records rather than
+DMA tags. That is consistent with the earlier finding that the packet's own
+control region is not a tag stream -- the game's per-class draw state is stored
+elsewhere and referenced, not inlined.