How Desecration works
Desecration is one of the few ways in PoE2 to choose a modifier instead of hoping for it.
- Apply a bone to a Rare item with an open prefix or suffix slot. The item gets an unrevealed desecrated modifier.
- Reveal it. The game shows three options; you keep one.
- If none fits, remove the result and try again. Omen of Light makes your next Orb of Annulment remove only desecrated modifiers, so the rest of the item is safe.
An item holds one desecrated modifier at a time: no further bone can be applied while one is on it. Plan which slot you'll fill with it.
Where the three options come from
One bone adds one modifier. The three rows in the window are candidates for that one place. You keep one; the other two are gone.
On a ring, all three are prefixes or all three are suffixes, depending on which slot the bone filled. If both sides have room, Omen of Sinistral Necromancy forces a prefix and Omen of Dextral Necromancy forces a suffix.
Each candidate comes from one of two pools:
- The ring's ordinary pool: the same modifiers the ring can roll from orbs, such as Fire Resistance. They roll tiers the normal way. The 41–45% elemental resistances seen on desecrated rings are top tiers of this pool, not a special roll.
- The Desecration-only pool: modifiers you cannot get any other way, such as Fire and Chaos Resistance. Each belongs to one of the Abyssal lords: Amanamu, Kurgal or Ulaman.
"Desecrated" means two different things. Desecration-only describes where a modifier comes from: its pool. A desecrated modifier is any modifier you added through the Well of Souls, whichever pool it came from.
Pick plain Fire Resistance in the window and it is still a desecrated modifier: Omen of Light can remove it, and it takes the one desecrated slot the item has.
In every one of the 563 windows we recorded, at least one row was Desecration-only. That is what we observed. Our model of it: one row is reserved for a Desecration-only mod, and each of the other two is Desecration-only about 8% of the time and ordinary otherwise. So the usual window is one Desecration-only candidate and two ordinary ones.
This does not promise any specific modifier. Which Desecration-only mod you get is a roll, covered below. The row it lands in means nothing either: it was top, middle and bottom about equally often (173 / 154 / 153 of 480 windows with a single Desecration-only row).
Can the mod appear on your ring?
Before any odds: some modifiers cannot appear in your window at all, and that depends on what the ring already has. The game sorts modifiers into groups, and an item can hold a group only once.
- Groups already on the item are not offered. Have a cold resistance suffix? No cold resistance option will appear.
- Attribute hybrids are blocked by either attribute. If the ring already has Dexterity, the hybrid Dexterity and Intelligence will not appear. We checked live trade listings: 101 items had desecrated Dexterity and Intelligence; zero also had plain Dexterity or plain Intelligence, while 13 had Strength.
- Resistance hybrids are not blocked. Fire and Chaos Resistance is its own group, so it appears next to plain Fire Resistance as often as next to Cold or Lightning (102 / 94 / 108 listings).
- No group appears twice in one window. In 563 windows, not one repeat. The ignite, freeze and shock prefixes share one group, so at most one of them is offered at a time.
The rule follows the game's internal groups, not the wording of the modifier. Defence hybrids such as % Evasion and Energy Shield coexist with plain % defences for the same reason.
So, can Fire and Chaos Resistance appear on your ring? Yes, if the ring has an open suffix slot. A plain Fire or Chaos Resistance suffix does not block it.
When a mod is blocked on your ring, its chance is zero and every other mod of that pool becomes a little more likely. This is why the chances below can differ for your ring.
On rings with a Preserved Collarbone, item level did not gate Desecration-only mods
All ring Desecration-only mods list required level 65, yet we saw them on rings of item level 22, 49 and 64 using a Preserved bone. A cheap low-level base works for hunting one. Ordinary options still respect item level.
How often will you see it?
Take one target: Fire and Chaos Resistance on a ring, with the bone filling a suffix.
About 1 window in 10. With Omen of Abyssal Echoes, which shows a second window: about 1 bone in 5.
The number behind it is the mod's weight, 8.4%. It is our estimate of how often the reserved row of the model is this mod, out of all Desecration-only suffixes. Weights for every mod are in the tables below.
The calculation, step by step
- The reserved row is this mod 8.4% of the time.
- Each of the other two rows is Desecration-only 7.5% of the time on the suffix side. Together they add 2 × 7.5% × 8.4% ≈ 1.3%.
- Per window: about 9.7%.
- With Echoes you see two windows: 1 − (1 − 0.097)² ≈ 18.5% per bone.
- That is about 5 desecrations on average (1 / 0.185 ≈ 5.4).
The same steps work for any Desecration-only mod with weight w:
P(screen) ≈ w × (1 + 2 × 0.08)
With Omen of Abyssal Echoes you see two screens: P ≈ 1 − (1 − P(screen))².
Expected number of desecrations ≈ 1 / P.
| Target on a ring | Weight | Per screen | With Echoes | Desecrations on average (with Echoes) |
|---|---|---|---|---|
| Fire and Chaos Resistance | 8.4% | ~9.7% | ~18.5% | ~5 |
| Strength and Dexterity | 11.8% | ~13.6% | ~25% | ~4 |
| Skill speed | 4.5% | ~5.2% | ~10% | ~10 |
| Cooldown recovery rate | 1.2% | ~1.4% | ~2.8% | ~36 |
The formula ignores groups blocked by your item. If you already hold a blocked group, that mod's chance is zero; the others get slightly more likely. Our solver accounts for this exactly.
A full simulation of the ring below, 20,000 runs with a Preserved Collarbone and Echoes, hit on 18.7% of bones, close to the formula's 18.5%. Spamming Chaos Orbs for the same mod is impossible: it is Desecration-only.
See the whole loop in the Constructor
We built this exact craft as a ready loop: a Gold Ring with life and mana prefixes and two resistance suffixes. Both sides have room, so Omen of Dextral Necromancy forces the suffix. Each cycle uses a Preserved Collarbone with that omen and Omen of Abyssal Echoes. A miss is removed with Omen of Light and an Orb of Annulment, so the other four mods are safe.
The Constructor simulates it 20,000 times and breaks the cost down by operation. On 13 November 2026 in Forbidden Rites the loop averaged ~50 divines, and ~49 of them went on removing misses. The bone and both omens cost about 1 divine in total. The misses are what you pay for, not the bones.
Open the loop in Constructor →
Pick your league in the top bar to use its prices. Change the bone, drop an omen or swap the target mod, then press Run to compare.
Hunting an ordinary mod through the reveal works differently, and the best bone depends on how narrow the target is. That is a separate short guide: when an Ancient bone is cheaper than a Preserved one.
in every window
is Desecration-only
in 563 screens
Bones: which one to use
The bone type decides which items it works on. The quality (Gnawed, Preserved, Ancient) sets the level range of the modifiers it can roll.
| Bone | Works on |
|---|---|
| Jawbone | Rare weapon or quiver |
| Rib | Rare armour |
| Collarbone | Rare amulet, ring or belt |
| Preserved Cranium | Rare jewel |
| Preserved Vertebrae | Rare waystone |
| Quality | Modifier level range | When to use |
|---|---|---|
| Gnawed | 0 – min(item level, 64) | Only on items up to item level 64. We have not measured whether it offers Desecration-only mods. |
| Preserved | 0 – item level | The default. All our measurements use it. |
| Ancient | 40 – item level | Cuts low tiers out of the ordinary options. Cheaper when the cost of the misses it saves outweighs the extra cost of the bones: see the ring example. |
Omens that change Desecration
| Omen | Effect | When it pays |
|---|---|---|
| Omen of Abyssal Echoes | Reroll the reveal options once | Almost always when hunting one mod: nearly doubles the chance per bone. |
| Omen of Sinistral Necromancy | Desecration adds only a prefix | When your target is a prefix and both sides have room. |
| Omen of Dextral Necromancy | Desecration adds only a suffix | Same, for suffixes. |
| Omen of Light | Next Orb of Annulment removes only desecrated mods | Cleaning up a failed reveal without risking the rest of the item. |
| Omen of the Liege / Blackblooded / Sovereign | Guarantee a random Amanamu / Kurgal / Ulaman mod (weapons and jewellery) | When your target belongs to one lord. Not measured by us. |
| Omen of Putrefaction | Replace all mods with up to 6 unrevealed ones and corrupt the item | A gamble, not a controlled craft. |
Removing misses is often the most expensive part of a desecration loop: in our ring loop it was about 98% of the cost. Omen of Light is not automatic. When the only removable mods are the unwanted desecrated one and one useful mod, a plain annul is a 50/50. Compare the omen's price with the cost of rolling the useful mod again.
Desecration-only mods are not equally likely
Data-mining sites list desecrated modifiers without usable weights. Our counts reject equal weights clearly on both sides. Below are relative weights for rings: when a row is Desecration-only, the chance that it is a given mod.
Ring prefixes
| Modifier | Lord | Weight |
|---|---|---|
| Remnant effect | Amanamu | 19.4% |
| Minion damage if you've hit recently | Amanamu | 16.0% |
| Spell damage while Energy Shield is full | Kurgal | 14.9% |
| Shock magnitude (Frenzy Charge) | Ulaman | 14.0% |
| Freeze buildup (Power Charge) | Kurgal | 13.6% |
| Attack damage while on Low Life | Ulaman | 12.4% |
| Ignite magnitude (Endurance Charge) | Amanamu | 9.8% |
Ring suffixes
| Modifier | Lord | Weight |
|---|---|---|
| Strength and Dexterity | Ulaman | 11.8% |
| Dexterity and Intelligence | Kurgal | 10.3% |
| Fire and Chaos Resistance | Amanamu | 8.4% |
| Remnant collection range | Amanamu | 7.9% |
| Skill effect duration | Amanamu | 7.9% |
| Lightning and Chaos Resistance | Ulaman | 7.2% |
| Cold and Chaos Resistance | Kurgal | 6.9% |
| Recover % of maximum Life on kill | Ulaman | 6.9% |
| Strength and Intelligence | Amanamu | 6.5% |
| Mana leech | Kurgal | 6.4% |
| Life leech | Amanamu | 6.2% |
| Recover % of maximum Mana on kill | Kurgal | 5.9% |
| Skill speed | Ulaman | 4.5% |
| Exposure effect | Kurgal | 1.9% |
| Cooldown recovery rate | Kurgal | 1.2% |
Exposure effect and cooldown recovery are 4–10× rarer than the other suffixes. Budget for dozens of reveals before chasing them, or buy them already rolled.
Weights for small samples carry a few percentage points of error, and these are ring weights. Weapon and armour desecrated mods not seen on rings are not measured; our solver gives them the average weight of their side.
How we measured it
The raw counts were first shared on Reddit: Part 1, how many reveal options are desecrated and Part 2, how often each desecrated mod appears. This guide collects the results and adds what they mean for crafting.
We prepared 573 cheap rings, applied Preserved Collarbones and recorded every reveal screen. Each side was forced by filling the opposite side, so we know which pool each screen came from. Screens were read from screenshots and matched against all 52 ring modifier templates.
- 563 valid screens, 1,689 options: 236 prefix and 327 suffix screens. Invalid cases were removed.
- Calibration: ordinary options on the free slots match Craft of Exile weights (correlation 0.92 prefix, 0.79 suffix), so the data is trustworthy.
- Weights are maximum-likelihood estimates that account for blocked groups. Attribute hybrids are blocked on about a third of screens, so their true weights are 29–43% above raw counts.
- Exclusion rules were checked independently on live trade listings.
Why one row looks reserved: the shared-pool model does not fit
A common belief is that Desecration-only and ordinary mods share one weighted pool. If they did, about 23% of windows would have no Desecration-only row at all. We recorded 563 windows and found zero of them.
A simple model fits these observations: one row is reserved for a Desecration-only modifier, and each of the other two is Desecration-only with a small fixed chance, otherwise ordinary. This is a model of observed data, not an official description of the mechanic.
| Desecration-only rows in a window | 0 | 1 | 2 | 3 | Chance per other row |
|---|---|---|---|---|---|
| Prefix side · 236 screens | 0 | 198 | 37 | 1 | 8.3% |
| Suffix side · 327 screens | 0 | 282 | 41 | 4 | 7.5% |
The Desecration-only share was also the same on both sides (38.8% vs 38.3%), even though the prefix side has 7 such mods and the suffix side 15. A shared weighted pool would make these shares differ, so the chance does not depend on the ordinary pool.
Two practical consequences:
- Hunting a Desecration-only mod is reliable. Every reveal is a real roll at it, whatever ordinary mods the item can still take.
- Hunting an ordinary mod costs more than it looks. Only two of the three rows can be ordinary, so each reveal gives fewer chances at it. In our ring model this made a reveal loop about 14% more expensive than a three-ordinary-option model predicted.
What is not measured yet
Gnawed bones, the lord-guarantee omens, and weapon or armour weights. The free-slot chance comes from 37–45 events per side, so it is about ±2 percentage points. Measurements were made in July 2026 (Runes of Aldur league). Recheck after a patch changes Desecration.
FAQ
Is a desecrated modifier guaranteed on reveal?
In all 563 ring reveals we recorded, at least one option was Desecration-only. A model with one reserved option and about 8% for each of the other two fits these observations. Which mod appears is still a roll.
Does item level limit desecrated modifiers?
Not on rings with a Preserved Collarbone: Desecration-only options appeared on item level 22, 49 and 64 rings. Other items and Gnawed bones were not measured.
Does the position of an option in the window matter?
No. The Desecration-only option lands in each position about equally often.
Is Omen of Abyssal Echoes worth it?
When you hunt one specific mod, almost always: it roughly doubles your chance per bone. A mod at ~10% per screen rises to ~19%.
Can the same modifier group appear twice?
No, and groups already on the item are not offered. Attribute hybrids are blocked by either attribute; resistance hybrids are not.
Let the solver count your reveals
The Advisor and Constructor use exactly this model: the reserved row, the measured weights, group exclusion, Echoes and Omen of Light. Paste your item to see whether desecrating is the cheapest next step, or build a desecration loop to see its expected cost.
Desecration in practice: when an Ancient bone is cheaper than a Preserved one, and Time-Lost Sapphire for Varashta minions uses desecration to land the fifth mod, and needs eleven reveals in one recorded attempt.