Demeterra watched the war from the root-throne.

The throne was not a seat — it was a living structure, a network of root-tendrils that rose from the earth of her divine sanctuary and wrapped around her physical form like the fingers of a devoted congregation. Each root connected to the Growth domain’s territorial awareness — the sensory network that extended across every territorial grid she controlled, feeding her the pulse of soil composition, moisture content, crop health, and the movement of every organism heavier than a field mouse within her divine borders.

Through the war, the roots carried additional data. Troop movements. Supply column velocities. Casualty figures transmitted through the domain’s organic relay from her commanders on the front lines.

Through the roots, she assembled the picture. Day twelve.

The Ashwall front: contested. Three breaches — western, central, eastern — each one clawed open by different Accord forces. Durnok’s Crushist cavalry had taken forty percent casualties in the opening engagement, above projections. The Frogman swamp bypass was proceeding at seventy percent of the original timeline — slower than planned, still viable. Thalveris’s engineering column was the wound that wouldn’t close: six of twelve Siege Shapers killed by a single unknown individual asset. The entire engineering deployment pushed back to Day 15 at the earliest.

Pale Coast: on schedule. Sylvaen’s naval bombardment of Tidewatch ongoing. The fishfolk defence was concentrated but underequipped for a sustained naval siege. Seylith’s divine attention divided between coastal defence and Covenant coordination — exactly where the Accord needed it: split.

Frostmarch: behind schedule. Morglith’s construct vanguard had suffered eighty percent losses in Whitefell Pass — environmental defence, the terrain itself weaponised by the Howlist pack tactics. Blightkin infantry still operational but under sustained harassment. Morglith’s own assessment: the northern front was a containment action, not a breakthrough.

Overall: Phase One objectives partially achieved. Phase Two timeline delayed. The Iron Sovereign’s military infrastructure had exceeded intelligence projections.

Demeterra absorbed the assessment with a patience that 253 years of existence had cultivated into a survival mechanism.

The war was not going well. But the war was going as she had expected it to go in her most conservative planning scenario — the scenario she had labeled *Iron Ceiling*, in which Zephyr’s military preparations proved superior to her intelligence estimates and the opening phase produced friction rather than breakthrough.

She had built the Green Accord for friction. Other coalitions — the ones that had formed and collapsed across Aerthys’s divine landscape over the centuries — were designed for overwhelming force: assemble enough power, crush the enemy in a single offensive, divide the spoils. Those coalitions broke when the single offensive failed, because they had no strategy for Phase Two, no institutional structure for sustained conflict, no mechanism for managing seven gods’ competing priorities when the war extended past the first month.

The Green Accord was designed for attrition.

***

Demeterra opened the Divine Communion — the system’s standard god-to-god communication channel — and pulled all six Accord gods into a simultaneous connection.

Any god could open a communion with any other god. Universal feature, not a personal power. But managing a six-way communion required a host, and Demeterra had hosted every Accord-wide communion for forty years — setting the channel’s parameters, moderating its flow, deciding who spoke and when.

Five presences entered the channel.

Gorvahn first — always first. The Mire Lord’s consciousness was thick, patient, the swamp-weight of a god who measured timescales in seasonal floods and who found haste philosophically offensive. His Frogman armies were the Accord’s most reliable infantry, and his loyalty to Demeterra was absolute — not from affection but from economic dependency. The Great Mire’s survival required Demeterra’s agricultural trade. Without it, Gorvahn’s population starved. Dependency produced loyalty more durable than any oath.

My forces in the marsh bypass are at sixty-seven percent original strength, Gorvahn reported. The enemy sealed one corridor with reserves — heavy infantry, competent commander, a war-construct with multiple attacks. Our advance continues through secondary routes. Progress is slower but sustainable. I have taken no losses I cannot replace within thirty days from rear-area reinforcement.

Durnok next — the Earthbreaker, whose patience was not philosophical but tactical. His Crushist cavalry had been mauled at the Ashwall’s central breach — the kill-zone geometry that Demeterra’s intelligence had underestimated and that had produced forty-percent casualties in the opening assault.

The cavalry losses are recoverable, Durnok stated. His consciousness carried the granite weight of the Earth domain — solid, heavy, slow to shift. The kill-zone geometry was unexpected. Their defensive engineering at the breach points exceeds anything we’ve encountered in previous wars. The trench networks, the prepositioned stonesteel obstacles, the fire-channeling — these are the innovations of a god who has been preparing for exactly this attack for years.

He has been preparing for decades, Demeterra corrected. The Iron Sovereign does not react to threats. He prepares for threats before they materialize — that is his entire military philosophy. Pre-positioned advantage, always. Do not express surprise at his preparations. Express surprise only if you find a position he has NOT prepared.

Thalveris entered the channel — the Fortress God, the coalition’s engineer, whose Fortification domain enabled the systematic demolition and reconstruction of fortifications at speeds that conventional armies couldn’t match.

My engineering column has been compromised, Thalveris said. Six Siege Shapers killed. A single attacker — unidentified, operating solo, penetrated the escort perimeter and killed them in under thirty seconds. A pause. In under thirty seconds, Demeterra. No conventional warrior can do that to six Fortification-blessed priests in protective custody.

The Sword Saint, Demeterra said. The individual asset that our intelligence identified before the war. Kael Verenthis. Unaffiliated, unranked by the Sovereign’s official military, and operating outside the normal command structure. Our pre-war assessment flagged him as exceptional. The assessment was accurate.

One man delayed my entire engineering deployment by a minimum of three days.

Yes. One man is worth three days. Against a different god, that imbalance would be exploitable — find the man, kill the man, eliminate the delay. Against the Iron Sovereign, the man is a symptom. The defensive system would have produced another obstacle if the Sword Saint hadn’t existed. The man is the system’s expression, not its cause.

Sylvaen joined — the Tidecaller, whose Currents domain forces operated as the Accord’s naval arm on the Pale Coast front. Her consciousness was thinner than the others — scattered across multiple operational theatres, the Tidalist naval forces dividing her attention between the Tidewatch bombardment and the coastal siege operations.

Tidewatch bombardment is proceeding, Sylvaen reported. The fishfolk defenders are consolidated around the harbour district. Our naval forces control the approaches. But the fishfolk are competent sailors and the harbour’s defensive architecture is stronger than assessed. We’re degrading their capacity rather than overrunning it.

Kreth entered last. The Scavenger — the Accord’s weakest god, the intelligence specialist, the Voice-domain deity whose contribution was espionage rather than military force.

Kreth’s presence was the thinnest of all — deliberately so. The Scavenger god maintained a minimal divine signature in all communications, his Voice domain specialty making him naturally difficult to read even through friendly channels.

Intelligence update, Kreth said. The enemy’s internal communication architecture is more sophisticated than pre-war estimates. Their divine relay system — the Sovereign’s multi-domain communication network — transmits intelligence between fronts with minimal delay. The Frostmarch commanders know what’s happening at the Ashwall within hours. The Pale Coast defenders coordinate with the central command in near-real-time. We are fighting an enemy who can see all four fronts simultaneously.

We can also see all four fronts simultaneously, Demeterra noted.

We can. Through the same system communion. But our coordination relies on dead-drops and encrypted mortal channels for operational details. They coordinate through a divine infrastructure — temples, priests, relay networks — that was designed for exactly this purpose by a god who thinks like a systems architect. The comparison is unfavorable.

***

Demeterra dismissed the channel and sat with the assessment.

The war’s first twelve days had produced a result that was neither victory nor defeat — it was measurement. The Accord had measured the Iron Sovereign’s defenses and found them superior to intelligence projections. The Sovereign had measured the Accord’s offensive capacity and found it persistent but not overwhelming. Both sides had spent twelve days calibrating their models against reality, and reality had proven — as it always did — that models were approximations and wars were not models.

The critical variable was time.

Demeterra’s war plan was built on a thirty-day operational timeline — ambitious in scope, aggressive in Phase allocation, and designed to produce maximum pressure within a window that the Accord’s logistical infrastructure could sustain. Beyond thirty days, the Accord’s supply chain — which operated across seven gods’ territories with varying infrastructure quality and multiple political complications — would begin degrading. Beyond forty-five days, the degradation would become unsustainable.

The Iron Sovereign’s sustainability was the question that Demeterra’s entire strategic calculation rested on.

She ran the analysis that mattered — the one the reports didn’t contain.

The Iron Sovereign’s industrial capacity was superior — the stonesteel economy, the centralised forge production, the infrastructure that turned raw material into equipped soldiers faster than any other god on the continent. His supply chain was integrated under single command authority with a developed road network. His manpower reserves were unknown — her best estimate put another fifteen to twenty percent beyond the currently deployed forces. His divine reserves were a Rank 7 FP pool, estimated at three to five million. And the three confirmed Heroes had not been deployed.

The assessment was simple: the enemy could sustain defensive operations for sixty to ninety days at current intensity. The Accord could not win an attrition war. The Accord had to force a decisive engagement before Day 30.

Three Heroes. Not deployed.

This was the variable that kept Demeterra awake — not physically, gods didn’t sleep, but in the attentional sense. Three confirmed Heroes. Krug the First Forge, and two others whose identities Kreth’s intelligence had been unable to determine with certainty. Heroes were divine-tier combatants whose battlefield deployment could eliminate armies, shatter divine constructs, and — in extremity — kill the gods who commanded those armies.

Demeterra had survived one war against Zephyr when he was Rank 3 with no Heroes. She had weathered two more when he was Rank 5 with one. Now he was Rank 7 with three. The power differential between then and now was exponential — a Rank 7 god with three Heroes was, in Demeterra’s assessment, capable of defeating any two Accord gods simultaneously in a direct divine engagement.

The Accord’s advantage was not power. It was architecture. Seven gods attacking on four fronts, forcing the Sovereign to distribute his attention, his reserves, his Heroes across a strategic landscape that was too wide for any single consciousness to optimize. Even a god who thought like a systems architect couldn’t optimise four simultaneous campaigns without degradation somewhere.

The question was where the degradation would appear. And whether Demeterra could exploit it before the Accord’s own degradation made the attack unsustainable.

She reached through the root-network to her personal forces — the 130,000 Rootist ground troops, the agricultural reserves, the faith infrastructure that 400,000 believers sustained. Her FP reserves were adequate: 1.8 million, enough for sustained divine operations including one Descent if the war demanded it.

The Descent was the card she held in reserve. A Rank 6 god’s Descent — the projection of divine consciousness into physical form on the battlefield — was the most devastating act a god could perform. At her rank, she could sustain physical form for approximately four minutes. Four minutes of divine-scale power: earth reshaping, Growth domain manipulation at continental intensity, the ability to destroy fortifications, shatter armies, and directly challenge the Sovereign’s defensive architecture.

Four minutes. The war’s thirty-day timeline might come down to whether those four minutes were spent in the right place at the right time against the right target.

Two hundred and fifty-three years. Twenty-four gods absorbed, defeated, or outlasted. Three wars against the Iron Sovereign — each one ending in stalemate because I was not strong enough, not prepared enough, not willing enough to commit everything.

This time, everything is committed. The Green Accord is everything I have. Every alliance, every vassal, every reserve, every favour, every threat. If this fails, there is no fourth war. If this fails, the Iron Sovereign will be strong enough that no coalition on Aerthys can challenge him.

This is the last window. And the window is thirty days wide.