EMERGENCY ALERT: Stay Away From the Ogeechee River Corridor.

The seven-fifteen pull at Station Seven takes me four minutes on a good day, three if the access gate hasn't rusted shut overnight. That morning it swung open clean, so I had the sample bottle uncapped and dipped below the surface before the sun had fully cleared the tree line to the east.

I've worked this stretch of Highway 17 for three years. Twelve miles, eleven active stations, one decommissioned platform at what used to be Station Four before a storm surge took out the access stairs back in the spring. Every morning, Monday through Friday, I run the route north to south and back, pulling samples, logging turbidity readings, checking instrument calibrations on the ones that have sensors. The work is quiet and it is mine and I have come to rely on both of those things in ways I don't examine too closely.

The water along this corridor runs tea-brown from October through February. Tannins out of the marsh grass and the decaying cordgrass beds, organic acids bleeding down through the sediment. It's the color of weak coffee left in a pot too long. I've photographed it a thousand times for the state database, filed it under normal seasonal variation, moved on. Nobody in Atlanta flags it. Nobody in Atlanta looks at much of anything I send, which suits me fine. The work gets done either way.

So when I lifted the sample bottle out of the water that morning, I noticed the color before I'd even straightened up.

It wasn't brown.

I stood there on the platform with the bottle at chest height, water still dripping off the outside of the glass, and looked at it. Dark. Deeper than the tannin stain I expected. There was a purple cast to it, not vivid, not something you'd catch in a photograph without the right light, but unmistakable in person. I tilted the bottle toward the sky, which was pale and overcast the way October mornings get down here, and the color shifted. An iridescent sheen across the surface, blue-green at one angle, then back to that dark purple as I rotated it.

I set it on the platform railing and looked at the water below. It looked normal from above. The surface had the usual chop from a northeast wind, the marsh grass along the bank was the same dried gold it gets in fall. Nothing floating, no film on the surface I could see from where I stood.

I picked the bottle back up.

My thermos was in the truck. I'd been three stations in already, and I'd been running on that coffee and a gas station biscuit since six-thirty, and part of my brain wanted to tell me I was tired and reading the color wrong. That part of my brain was wrong. I've pulled enough samples to know what I'm looking at.

I set the bottle against the white foam insert in my sample case and pulled out my field notebook. The platform number is stamped into the concrete on the south corner: Seven. I wrote the time, seven-nineteen, the air temp from the thermometer clipped to my vest strap, fifty-one degrees, the wind direction, the GPS coordinates I already had memorized but logged anyway. Then I wrote: color anomaly, purple-iridescent, no surface film visible, no odor noted. I leaned down and smelled the open case. Nothing unusual. No sulfur, no petroleum, no algal smell.

I took my phone out and photographed the bottle against the white foam, then against the sky, then against the railing so the water was visible in the background. Seven shots. I got down on one knee and photographed the water surface directly, which gave me nothing useful but felt like the right thing to do.

Three years of this work and I have never opened a training manual that described what I was holding. The state certification course covered petroleum sheens, algal blooms, coliform indicators, turbidity spikes from runoff events, sediment disturbance from boat traffic. There's a whole module on red tide identification, which doesn't apply here geographically but they include it anyway. None of it matched. Red tide is rust-brown or orange-red. Algal blooms green out the water or leave a surface scum you can smell from the access road. A petroleum sheen sits on top and breaks into rings when you disturb it.

This was throughout the sample. Uniform. Not a surface phenomenon.

I stood there another minute. The wind came off the marsh and the portable radio on my belt clicked once with static the way it does when a truck passes on the highway a quarter mile west. Everything else was the normal sounds of the place — red-winged blackbirds working the grass edge, a heron somewhere further north, the low hiss of the water moving under the platform.

I capped the bottle.

I carried it back to the truck and set it in the rack behind the driver's seat, which is a repurposed wire shelf I bolted to the cab wall years ago to keep samples from rolling around. I have a label maker in the glove box. I printed a label: ST7 / 10-14 / 0719 / ANOMALY and stuck it on the bottle. Then I sat in the driver's seat for a moment with the door open.

Station Nine was four miles north. I had eight more stations to run before noon.

I pulled the door shut, turned the key, and headed north on Seventeen with the windows down and the anomaly sitting in its rack behind me, labeled and logged and completely without explanation.

Station Nine sits off a gravel pull-through about forty yards from the highway, down a short access track that floods out twice a year when the tidal creeks push back hard enough. I'd driven it so many times I didn't have to think about the turn. Just checked the mirror, eased off the pavement, and felt the truck drop into the ruts.

I smelled it before I parked. Not sulfur, not the low-tide mud smell I was used to, not anything I could name right away. Just wrong. The kind of wrong that registers in your chest before your brain catches up.

I got out and walked the ten feet to the creek bank.

There were hundreds of them. Probably more. Drum and mullet mostly, from what I could tell, though some of the larger ones had turned pale enough that the species markings were gone. They covered the surface in a dense silver layer that started at the bank edge and stretched out around the bend to the north until the creek curved out of sight behind the spartina. The ones closest to me were stacked two and three deep at the waterline where the bank curved in, pressing against each other in the stillness.

I stood there for a second. Just stood there.

The creek wasn't moving. That was the thing that took another moment to register. The tide was running — I'd checked the table that morning, flood tide through roughly ten-thirty — but the surface was flat. No current ripple along the grass stems. No drift on the fish. They weren't moving downstream or up. They were just sitting there on dead water like someone had poured them out of a bucket and the whole system had locked up around them.

I went back to the truck and pulled the YSI meter from the side pocket of the equipment bag. It's a handheld dissolved oxygen probe, yellow casing, about the length of my forearm. I've used the same model for three years. I know its quirks — it reads about half a unit high in cold water below forty-five degrees, it needs a full sixty-second equilibration before the number stabilizes, and the display sometimes flickers if the battery connection is loose. I know all of that. I'm telling you this because what it did next had nothing to do with any of those quirks.

I switched it on, waited for the initialization screen to clear, walked back to the bank, and lowered the probe into the water near the edge of the fish mass.

The reading started moving immediately. Not down toward zero, the way you'd expect with a severe hypoxic event. It went past zero. The display showed a negative value — negative two point four — and then the unit started cycling. Error code E-seven. Then E-twelve. Then a sequence I'd never seen, a three-digit code that wasn't in the field manual I kept in the glove box. I pulled the probe up, let it equilibrate in the air for thirty seconds, and put it back in. Same result. Negative. Error codes cycling in a sequence that had no business existing on a meter that was supposed to read dissolved oxygen in milligrams per liter.

I took photos of the display. Five of them, making sure at least two showed the error codes clearly. Then I walked the bank north until the fish layer ended at a line of submerged roots about sixty feet up, photographed that too, and came back.

There was no odor from the water itself. No decomp smell from the fish, either, which didn't make sense given the volume of them. That registered as its own separate thing I didn't have a category for.

I called Brandon from the bank.

He picked up on the second ring, which meant he was already at his desk in Brunswick. I gave him everything in order: Station Seven's purple-iridescent sample from that morning, Station Nine's fish kill covering the creek surface and extending beyond visible range, the DO meter returning negative values and error codes outside the instrument's documented range, and the creek surface showing no tidal movement despite confirmed flood conditions. I didn't editorialize. I read him the numbers I'd written in the field notebook while I was talking.

He was quiet for a moment after I finished.

Then he told me to stay put. Get the sample collection done, keep the meter in the water at intervals and log every reading regardless of what the display showed, photograph the fish kill from multiple angles including from the access track, and wait for the secondary team out of Brunswick. Three hours, he said. They'd come straight up Highway 17 and meet me at the Nine access pull-through.

I said I understood.

He said it again. Stay at Nine, Dalton. Don't move the truck.

I said yes.

We hung up.

I stood at the bank for another minute looking at the fish. The ones near the edge had started to develop a slight iridescence along their lateral lines, a faint blue-green color that ran the length of each body. Not all of them. Just the ones closest to the bank. I photographed that too and wrote it down.

Then I collected the water sample, labeled it the same way I'd labeled the Station Seven bottle, and carried it back to the rack in the cab.

I sat in the driver's seat with the door open. The access track ran straight back to the highway. Brandon had told me to stay. The secondary team was three hours out. The right move, procedurally, was to document, hold the scene, and wait.

I looked at my notebook. Station Ten was two miles north. It was a dry-season creek bed most of October — I'd logged it as near-zero flow three times in the last month — which meant whatever had moved through Station Nine's water and left those fish behind may have passed through Ten before it got here.

I closed the door and started the truck.

I pulled back up onto Highway 17 heading north, the sample rack rattling faintly behind me.

Station Ten sits at the end of a two-track access path that branches off Highway 17 through a stand of loblolly pine before opening onto a low bluff above a tidal creek. I'd been here three times in the past month. Every time, the channel had maybe four inches of slow water in it, sometimes less, enough to dip a sample bottle if you stretched down from the bank edge and held onto a root. October flow on this stretch was always minimal. I knew that going in.

What I did not know going in was that there would be no water at all.

I pulled up to the bluff edge and sat in the truck for a moment looking at it. The channel was exposed bank to bank — maybe twenty feet across — and the floor was cracked gray mud, the kind you see in August drought conditions, not mid-October with a flood tide supposed to be running two miles south. The cracks ran in irregular polygons across the flat, the edges curled slightly upward like something had dried it out fast. I've seen exposed creek beds before. They have a smell — the anaerobic stuff that gets locked in the sediment, hydrogen sulfide, the low-tide mud that coats the back of your throat. I know that smell well enough that I barely register it anymore.

This wasn't that.

It hit me when I opened the truck door. Sharp and chemical, almost medicinal, but not any specific chemical I could name. Not petroleum. Not solvent. Not anything from the runoff categories I'd been trained on. It was its own thing, and it was coming up off the mud flat in a way that made me not want to breathe through my nose. I pulled my forearm up across my face and held it there while I walked to the bank edge.

The channel floor was completely still. No water pooling in the cracks, no seep coming in from the banks, no sign anything had drained recently — no waterline stain on the exposed bank walls, no debris line, nothing to indicate a gradual recession. It was dry the way a thing is dry when something removed the water rather than when water left on its own. I don't know how else to put it. The distinction felt significant at the time and I couldn't explain why.

I got my phone out and started photographing. The mud flat. The bank cross-section. The cracked surface pattern close up. I was trying to document the chemistry smell by writing in my field notebook — "sharp, chemical, non-sulfur, non-petroleum, source unknown, concentration strong at bank edge, dissipates approximately ten feet back" — and that's when I looked up from the notebook and saw the tracks.

They were crossing the channel floor in a straight line from the near bank to the far bank. Circular impressions in the cracked mud, each one roughly eighteen inches across, with a central depression that went noticeably deeper than the surrounding contact area, and five radial grooves extending outward from the center like spokes. Clean-edged. Deliberate geometry. They weren't random pressure marks and they weren't the kind of thing that forms from water movement or gas venting. Something had made them by stepping there, and it had crossed the channel on a straight bearing without deviation.

I stood on the bank and looked at them for a while.

Three years of state certification coursework. Macroinvertebrate ID, sediment profiling, water chemistry, runoff event response. I've got a laminated track identification card in the truck that covers thirty-two native Georgia species. Nothing on that card has five radial grooves and an eighteen-inch diameter. I didn't need to check.

I photographed each impression individually, then stepped back and shot the full line of them crossing the channel floor. Seven tracks total, evenly spaced, each one pressed to the same depth. I counted them twice. The spacing between each impression was consistent enough that whatever made them had a regular stride, and the line they formed didn't curve or wander. It went straight from one bank to the other and climbed up into the cordgrass on the far side.

That's where the line ended. The cordgrass swallowed the last track about four feet up the opposite bank, and after that there was nothing to follow visually — just the dense stalks, dried gold from the season, packed close enough together that I couldn't see more than a foot or two into them.

I was still on my side of the channel. Twenty feet of cracked mud between me and the far bank, and I hadn't crossed it. I'm not sure I'd made a conscious decision about that. I'd just stayed where I was.

I followed the track line back with my eyes to the near bank — my side — and tried to find where the first impression started. The bank edge crumbled a little where it met the channel floor, and the mud there was disturbed enough that the entry point wasn't clean. But the first clear track was maybe three feet out from the bank base, and the angle of approach put the origin somewhere in the pine stand behind me. Behind my truck.

I turned around.

The access path was empty. Loblolly trunks, needle duff on the ground, the back end of my truck. Nothing moving. I stood there long enough that it started to feel like the kind of thing I was doing to reassure myself rather than because I expected to find anything.

When I turned back to the channel, the cordgrass on the far bank was moving.

Not a lot. Not a violent thrashing. Just a slow, continuous lateral displacement of stalks, the kind of movement that means something is pushing through at low height, below the seed heads. The disturbance was maybe four feet wide and it was tracking toward me — not across the channel, not parallel to the bank, but angling in my direction, following the bluff line from south to north, closing the distance to where I was standing.

There was no wind. The morning had been still since I left Station Nine, and the stalks on my side of the channel weren't moving at all. Just the ones over there, in that four-foot corridor, working methodically through the grass.

I watched it for maybe ten seconds. The movement didn't speed up or slow down. It had a pace to it, steady and even, and it covered ground.

I got back in the truck.

I didn't run. I walked back up the bank at a normal pace, which I think was more about not doing something stupid than about actually being calm, and I got in, pulled the door shut, and started the engine. I sat with it running and looked at the channel through the windshield. From the truck cab I couldn't see the far bank well — the bluff angle cut off the lower cordgrass — but the upper stalks were still, and whatever had been moving through them had either stopped or dropped below my sight line.

I put the truck in reverse and backed up the two-track to the highway without turning around.

Out on Highway 17 I sat with the engine idling and looked at my field notebook, open on the passenger seat. The last entry read: "ST10 — channel fully dry, bank to bank. Chemical odor, non-sulfur, non-petroleum. Seven circular impressions, approx. eighteen inches diameter, five radial grooves each, crossing channel floor on straight bearing, near bank to far bank. Evenly spaced. No matching species in field reference." Below that I'd written the time: eight fifty-one.

I picked up my phone and called Brandon.

He answered on the second ring again. I told him I was at Station Ten, not Station Nine, and before he could say anything I told him what I'd found — the dry channel, the chemical smell, the tracks, and the cordgrass moving with nothing to explain it. I read him the track dimensions directly from the notebook. Eighteen inches across, five radial grooves, seven impressions, straight-line crossing.

He was quiet for a moment.

Then he asked me to repeat the diameter. I did. He asked if I was certain about the radial count. I told him I'd photographed each one individually and counted twice.

Another pause. When he spoke again his voice had a different quality to it — not alarmed exactly, but slower, like he was choosing words more carefully than usual. He told me the secondary team was still on Highway 17 south of Brunswick. He told me to drive north to Station Eleven, get eyes on the site, and not to exit the vehicle until he called me back. He said he needed to make another call first.

I asked him who he was calling.

He said he'd explain when he reached me, and then he ended it.

I sat there a minute with the phone in my hand. Brandon Wheeler has been doing hydrology work in coastal Georgia for longer than I've been alive. In three years I had never heard him cut a call short to make another one. I put the truck in drive and headed north toward Richmond Hill Road.

The Richmond Hill Road bridge isn't much to look at from the highway — a low concrete span over a tidal creek, rusted iron framework underneath, the kind of crossing you go over forty times before you actually look at it. I knew Station Eleven was on the south bank, downstream of the bridge shadow. I'd been there enough times that pulling off onto the gravel shoulder was automatic.

What wasn't automatic was the DNR truck sitting sideways across the access road with the driver's door hanging open, engine still running.

I pulled up behind it and sat for a second. The truck was a late-model green-and-white, state plates, Georgia DNR insignia on the door. A man was standing twenty feet ahead of it, facing away from me, both hands at his sides. He wasn't moving. Just standing there looking up at the bridge understructure.

I got out. Gravel under my boots, wind off the creek, the engine of his truck idling low. I walked up alongside him and he turned his head toward me.

I didn't know him. Stocky, broad-chested, olive uniform, the agency patches on his shoulders. His face was the color of old concrete — not pale the way someone looks after bad news, but a deeper kind of drained, the kind that takes some minutes to arrive. He looked at me for about two seconds. Then he pointed underneath the bridge without saying anything.

I looked.

There was something attached to the iron crossbeams on the underside of the bridge. Not built there, not part of the structure. It was attached to the underside of the easternmost span, roughly centered between the two support piers, and it was big. About the size of a refrigerator lying on its side. It hung from the iron framework with what looked like narrow filaments anchoring it at six or seven points along the top, pressed up into the gap between two parallel beams.

The surface was translucent. Glistening. The texture wasn't uniform — there were folds and slight ridges across it, and the whole surface had a wet sheen even in the flat overcast light. Through the membrane I could see movement. Dark shapes, slow-moving, like shadows behind frosted glass — distinct enough to track but not enough to resolve into anything I could name. Two of them. Maybe three. Drifting.

I stood there for probably fifteen seconds before I remembered to breathe.

The iron understructure had some surface rust on it, the normal brown-orange oxidation you see on anything that spends its life over saltwater. The filaments holding the mass up were lighter in color than the tissue itself — almost white, slightly opaque. Where they contacted the iron beams there was no corrosion visible, no chemical staining. They just held.

Then one of the interior shapes moved toward the outer membrane. It pressed against it from inside, slowly, the way you'd press a hand flat against a shower curtain. The membrane deformed outward under the contact, held the impression, then very slightly rebounded as the pressure maintained. I could see the outline of it through the tissue.

It was a limb. Long, articulated at two or three points, tapering toward the end. At the terminal end the contact spread into five projections radiating outward from a central point. The same geometry. The same count. The same configuration I'd photographed in the creek mud at Station Ten two hours ago.

My field notebook was in my jacket pocket. I didn't reach for it.

The limb held its outline against the membrane for four or five seconds and then withdrew back into the interior. The membrane surface smoothed. The dark shapes inside continued their slow drift.

Then the mass contracted. One long, slow inward pull — the whole surface tightening, the folds pressing together, the filament anchors pulling taut. It lasted about three seconds and then released. One full rhythmic cycle.

After the contraction the interior shapes separated. One became two. Two became four. The new forms were smaller than the originals but moved the same way, drifting on whatever interior current was running through the tissue. Thirty seconds, maybe slightly under. Four distinct shadows where there had been one.

The man beside me hadn't moved.

I looked at the water under the bridge. The purple iridescence was there — same quality as the Station Seven sample I'd pulled that morning, same blue-green shift when the surface caught any angle of light. It ran under the bridge and extended maybe forty feet downstream before the creek curved and I lost the line of it. The tide was in. The water was moving the way water should move. Only the color was wrong.

I turned back to the man. His name tag read Armstrong. He was still watching the bridge understructure with that drained expression, jaw set, not blinking much.

"Jim Armstrong," I said, reading it off his chest.

He nodded once.

I told him my name, told him I was EPD corridor monitoring. He nodded again. His breathing was slow and deliberate, the kind of breathing a person does when they're working to keep it that way.

He reached for the radio clipped to his shoulder strap. His hand was steady, which I noted. He keyed it and started talking — location, description, request for supervisor contact. His voice was flat, factual, low. He got through the location and a basic physical description of what was on the bridge before the dispatcher's voice came back through the speaker.

I could hear it through my own open truck window fifteen feet back. The dispatcher asked him to repeat. Armstrong repeated it. The dispatcher asked him to confirm the location again, then asked if he needed medical assistance, then asked him to hold. There was a long pause and a second voice came on, also confused, asking Armstrong to clarify what he meant by attached to the underside. Armstrong clarified. The second voice went quiet for a moment and then said they were contacting his field supervisor.

Armstrong lowered the radio to his side and looked back at the bridge.

Above us, on the highway, a truck passed without slowing. The bridge didn't move.

Armstrong's radio went quiet for a stretch. Not off — I could still hear the carrier tone bleeding through the open window of my truck — but the dispatcher had stopped asking questions and started conferring with someone off-mic. Armstrong stood with the radio at his side, watching the bridge. I sat in the idling truck with both hands on the wheel and did nothing for a minute.

That's when it came together.

I'd been working north all morning. Station Seven, then Nine, then Ten, then here. Purple water at Seven. Dead creek at Nine. Dry channel and tracks at Ten. The mass under the bridge at Eleven. I'd been reading that as escalation — like I was closing in on something, following a contamination event upstream toward its source. That's the job. That's how the training runs. You trace anomalies back to point of origin, you identify the mechanism, you document and report. I'd been doing exactly that.

But the mass under the bridge wasn't a source. It was a left-behind thing.

The iridescence in the water below the bridge ran downstream, toward Nine and Seven — not up. I'd noted that when I was standing there but hadn't finished the thought. The trail ran south. Whatever had passed through this corridor had moved south to north, through Seven first, then Nine, then Ten, and then through here — and the mass on the bridge wasn't where it stopped. It was what it left when it kept going.

I thought about the tracks at Station Ten. Seven of them, perfectly spaced, straight line across the channel floor, entering from the south bank and exiting north. Clean impressions, consistent depth, identical geometry all the way across. That's not an animal checking the water and turning back. That's passage. The channel was dry because the water was already gone by the time I got there, and whatever had crossed it hadn't been slowed by that at all.

The dissolved oxygen at Nine had been negative. I'd been trying to fit that into a contamination model since eight in the morning — anoxic event, chemical discharge, runoff — and none of it held because negative DO doesn't exist in any natural framework I was trained on. But if the water at Nine wasn't depleted by something introduced into it, if the condition was a byproduct of something moving through it, that read differently. The meter hadn't been reading the water. It had been reading what was left after the water had been used for something.

I pulled the field notebook off the passenger seat and looked at the time stamps without opening it. Seven nineteen at Station Seven. Sometime around nine thirty at Nine. Eight fifty-one at Ten. It was past nine forty-five now, maybe pushing ten. The corridor runs twelve miles. Whatever I was trailing had covered at least eight miles of it before I reached the first station at dawn, and it hadn't stopped here.

North of Station Eleven is Richmond Hill proper, then the I-95 interchange, then the Bryan County line, then open marsh corridor all the way up toward the Ogeechee's main stem. I don't have stations past Eleven. My route ends here. Whatever's north of this bridge is outside my coverage and outside my dataset, and I had twelve labeled sample bottles in a rack behind my head that documented the southern half of a trail with nothing on the northern end.

Armstrong's radio crackled. The second dispatcher voice came back on and said they'd reached his field supervisor and were patching through. There was a brief tone, then a male voice I didn't recognize — older, flat Georgia cadence — asking Armstrong for a location confirmation and a status. Armstrong gave it to him clean. No extra language, no hedging. He said he was at Station Eleven under Richmond Hill Road, that there was an unidentified structure attached to the bridge understructure approximately the size of a large appliance, that it was anchored and appeared to contain mobile interior forms, and that an EPD corridor technician was on-site as well. The supervisor asked how long it had been there. Armstrong said he didn't know. The supervisor asked if it posed an immediate public safety hazard. Armstrong was quiet for two seconds, then said he couldn't determine that from his current position.

I set the notebook back down.

The supervisor told Armstrong to maintain visual and not approach the structure. Armstrong said understood. The supervisor said he was calling the regional office and would be back in contact within fifteen minutes. Armstrong lowered the radio and looked at the bridge again. He hadn't glanced back at my truck once since I'd been sitting in it.

Fifteen minutes. Brandon had told me he was making another call before he called me back, and that call had happened more than thirty minutes ago now. The secondary team from Brunswick was still over two hours out. Armstrong was covered — he had a supervisor inbound on the radio and dispatch holding. The mass on the bridge wasn't going anywhere; the filaments were anchored into the ironwork and the interior forms were drifting, not building toward anything I could read as urgent.

What was north of here had a two-hour head start at minimum, probably longer.

I put the truck in drive.

I didn't say anything to Armstrong. I don't know that it would have changed anything if I had — he was fixed on the bridge and locked into his radio chain, and telling him I was leaving wasn't going to give either of us information the other needed. I rolled forward off the gravel shoulder, checked the mirror once as I turned north onto Highway 17, and watched his figure get small behind me. He didn't move. The DNR truck was still sideways across the access road, engine still running, driver's door still open. It looked the same as when I'd arrived.

The road north was empty. Flat marsh on both sides, dried cordgrass, the overcast sky sitting low over the tree line. My two sample bottles were racked behind me. The DO meter was in the equipment bag on the floor of the passenger side. I had a full tank of gas and a thermos that had gone cold two hours ago.

I kept both hands on the wheel and drove.

Fort Argyle Road comes up fast on the left, and I pulled off before I'd consciously decided to. The intersection isn't much — a stop sign, a county road heading west into scrub pine, the marsh running flat on both sides of Seventeen under a sky that hadn't moved all morning. I put the truck in park and left the engine running.

Behind my head the sample rack held twelve bottles in their foam slots. Two of them were the problem bottles. The rest were routine, pulled before seven in the morning at stations that had shown me nothing unusual — turbidity within range, conductivity normal, the kind of readings that go into the database and disappear. I could see the rack in the rearview if I tilted it, but I didn't. I already knew what was in there.

I picked up my phone and called Brandon.

He answered on the second ring, same as always.

I told him I was parked at the Fort Argyle Road intersection on Seventeen. I told him I was no longer at Station Nine, where he'd instructed me to stay, and I didn't editorialize about that. I opened the field notebook on the passenger seat and read from it in sequence, station by station, starting with the seven-nineteen time entry at Station Seven and the purple-iridescent sample I'd pulled before the sun fully cleared the tree line. I gave him the color description, the uniformity through the sample volume, the absence of surface film, the absence of odor. I moved to Station Nine: the fish kill, the dissolved oxygen reading of negative two point four milligrams per liter, the three error codes including the undocumented one, the still water during confirmed flood tide, the blue-green iridescence developing along the lateral lines of fish nearest the bank. I told him the decomposition smell wasn't there. I told him the creek smell wasn't there either.

He didn't interrupt.

Station Ten: the dry channel, fully bank-to-bank across twenty feet with cracked gray mud showing upward-curled polygon edges consistent with August drought conditions. The chemical odor, sharp and medicinal, nothing from training that matched it, strongest at the bank edge and dissipating ten feet back. Seven tracks crossing the channel floor on a straight bearing, each eighteen inches in diameter with five radial grooves from a central depression, evenly spaced, identical depth, perfectly straight line of travel south to north. I told him the track identification card in my glove box covers thirty-two native species. I told him none of them were relevant.

I told him about the cordgrass moving on the far bank with no wind. Four feet of lateral displacement tracking along the bluff line toward my position. I told him I got back in the truck.

Then Station Eleven. The DNR officer, Armstrong. The mass under the bridge — refrigerator-sized, translucent membrane, anchored to the iron crossbeams at six or seven points by filaments pressed into the beam gaps. Interior shapes moving through the membrane. One of them extended a limb against the surface from inside: long, articulated, tapering, five projections at the terminal end radiating from a central point. Same geometry as the tracks at Ten. Same count. Then the full contraction, three seconds, and after it the interior forms divided — one to two, two to four, the four smaller than the originals. Purple iridescence in the water below the bridge running forty feet downstream before the creek curved and I lost the line.

I told him the iridescence runs downstream, not upstream. Toward Nine and Seven, not away from them. Whatever moved through the corridor came from the south and traveled north, and the mass at Eleven wasn't the end of it. It was a stop.

Brandon was quiet for a moment. Not the kind of quiet where someone is confused. The other kind.

He said my name once, not as a question.

I told him I was driving north. I didn't ask him if that was the right call. I gave him my phone's battery percentage — sixty-one percent — and told him I'd find a charger.

He said he'd call me back.

I said fine and ended the call.

The intersection sat empty in both directions. A red-tailed hawk was perched on the stop sign post and it watched me pull back onto Seventeen without moving. I left it there.

The Hyatt Regency on Bay Street sits on the riverfront bluff, and I'd driven past it dozens of times over three years without ever going inside. I parked in the deck off Barnard Street, pulled the equipment bag and my field notebook and my phone off the seat, and left the sample bottles in the rack. The truck was locked. The bottles weren't going anywhere.

The lobby was cool and dim after the flat gray light outside. I asked for a room. The woman at the desk gave me one on the fourth floor, river-facing, and I paid with the EPD procurement card I carry for field emergencies — equipment failure, extended monitoring events, unplanned overnight stays. I'd used it once before, three years ago, for a motel off I-95 when my truck blew a tire at eleven at night. I didn't think anyone in Atlanta would question this one, and if they did I'd deal with it then.

The room was clean and quiet and smelled like hotel laundry. I set the equipment bag on the luggage rack, plugged my phone into the outlet beside the desk, and then I opened the field notebook to the first entry from this morning and started laying everything out on the bed.

The photographs were on my phone — one hundred and fourteen of them by the time I'd counted through — and I went through them in sequence and typed a log of each batch into the notes app with timestamps pulled from the EXIF data. Station Seven: seven images, time-stamped between seven-nineteen and seven twenty-two. Station Nine: thirty-one images covering the fish kill, the DO meter display, the iridescent lateral lines, the kill boundary at the root line. Station Ten: forty-three images, track by track, the full line, close-up geometry, the dry channel cross-section, the chemical smell's approximate boundary marked by where I'd stopped walking. Station Eleven: thirty-three images, the mass on the bridge, the membrane deformation, the limb impression, the iridescent water below.

The field notebook had handwritten entries for all four stations. I transcribed them into a typed document, pulling the exact language I'd used in the field: "color anomaly, purple-iridescent, no surface film visible, no odor noted." "DO reading: negative 2.4 mg/L. Error codes: E-7, E-12, undocumented three-digit code." "Sharp, chemical, non-sulfur, non-petroleum, source unknown." I didn't edit any of it. I didn't need it to sound better than it did.

Chain of custody for the samples: two bottles with anomaly designations, Station Seven at seven-nineteen and Station Nine at approximately nine-thirty, both labeled with the station, date, time, and anomaly notation, both seated in the foam rack in the locked truck three levels below me in the parking deck. I noted the storage conditions, the ambient temperature outside — around fifty-three degrees Fahrenheit by mid-morning — and the fact that neither bottle had been opened after collection.

I spread it all across the bed and stood at the foot of it and looked at it.

Twelve miles of corridor, dawn to mid-morning, and the trail didn't terminate anywhere I'd been. The mass under the bridge was already past tense by the time Armstrong found it. Whatever produced those tracks at Ten had crossed a dry creek bed and kept moving. The iridescence at Eleven ran south because that was the direction of travel — behind it, not ahead.

It had a head start measured in hours. Maybe more. Station Seven had been the southernmost sign of it, and I'd been at Seven before sunrise.

I sat down on the edge of the bed beside the notebook and looked at the river through the window. The Savannah was wide and gray out there, running east toward the sound, carrying whatever it carried.

My phone showed sixty-eight percent now, charging. Brandon hadn't called back yet.

I left the documentation spread across the bed and waited.

The phone hit seventy-two percent while I was still sitting on the edge of the bed looking at the Savannah River. Wide and gray and running east the way it always does. I'd been watching it without thinking about it, which is how I knew I'd already decided.

I got up and pulled the desk chair around to face the bed and sat back down in front of the spread of documentation. Field notebook open to the first morning entry. The notes app on my phone with a hundred and fourteen images indexed by timestamp and location. The typed chain-of-custody log I'd built over the past hour, verbatim language from the notebook, nothing smoothed out or reworded. I read through it once more from Station Seven to Station Eleven, tracking the logic of the corridor sequence the same way I'd tracked water chemistry anomalies for three years — looking for gaps, inconsistencies, anything I'd logged in error.

There wasn't anything. It held together the same way it had held together since I'd worked it out sitting on the gravel shoulder at Fort Argyle Road.

I opened my email and started a new message to Brandon, cc'd to the EPD Brunswick regional office address and the HR administrative contact I had filed under a tab in my phone from when I'd updated my direct deposit two years back. The subject line read: Resignation — Dalton Henry, Corridor Monitoring Technician, Ogeechee Corridor Highway 17 District. I wrote it the same way I'd write a field report. Effective immediately. Three years of service. All samples collected on October fourteenth would be transferred to the state lab in Atlanta with full chain-of-custody documentation. The agency truck would be left in the Hyatt Regency parking structure on Barnard Street with keys at the front desk. I thanked Brandon specifically by name and title, because he'd been a straight shooter for three years and that warranted the sentence. I read the message twice, made one correction — I'd typed "station" when I meant "station platform" — and sent it.

It didn't feel like much. That was probably right.

The twelve sample bottles were in the rack bolted behind the driver's seat in the truck three levels down in the parking structure. I put the equipment bag over my shoulder, took the field notebook and phone, and rode the elevator to the lobby. The front desk clerk told me the business center was off the corridor past the restaurant, second door on the left, open twenty-four hours.

The chain-of-custody form was a Georgia EPD Sample Transfer document, which I had as a template on my phone because I used them often enough that I'd learned not to rely on having paper copies in the field. I printed four copies in the business center — one for my records, one for Atlanta, one for the shipping carrier, one for the truck file I'd leave in the glove box. Filled them out by hand at the small table next to the printer. Each of the twelve bottles listed by station, collection time, collector name, sample type, and anomaly status. The two designated anomaly bottles — Station Seven at 7:19 a.m. and Station Nine at approximately 9:30 a.m. — got a separate notation on the form: "Possible biological or chemical contamination event, nature unidentified, do not open without full PPE protocol, contact EPD Brunswick regional office before processing." I signed and dated all four copies.

Then I went down to the truck.

The parking structure smelled like exhaust and damp concrete. My truck was where I'd left it, the sample rack visible through the rear window glass — twelve white foam slots, twelve bottles upright. I unlocked it, pulled out the sample case from behind the seat, and checked every bottle. Labels intact. Caps secure. The two anomaly bottles were undisturbed, still at roughly ambient temperature. I set the case on the passenger seat, took the equipment bag from my shoulder and put it in the truck bed, and sat in the driver's seat for a minute with the door open.

The glove box had the field manual, the laminated track ID card with thirty-two native Georgia species, and the DO meter's field reference sheet. I took the track card out and looked at it. Nothing on it was close. I put it back and closed the glove box.

I drove the truck to a FedEx on Drayton Street, twelve minutes from the hotel. Packed all twelve bottles in a foam-lined shipping box I bought at the counter, sealed it with the chain-of-custody form taped to the top surface under clear tape, addressed it to the state environmental lab in Atlanta — the same address I'd been routing samples to for three years. Paid with the EPD procurement card. The clerk asked if the contents were fragile. I said they were liquid samples and needed to stay upright. She put "FRAGILE / THIS SIDE UP" stickers on four faces of the box and printed the receipt.

I stood at the counter and watched the box go onto the processing shelf behind her.

That was the last physical evidence I had. Whatever the Atlanta lab did with it — whether they catalogued it, shelved it, flagged it, or lost it in a backlog — was no longer mine to manage. The photographs were on my phone and in the typed log, and I'd email both to Brandon before I left the hotel. The field notebook I'd keep. That was mine.

I drove back to the hotel and parked in the same level I'd used before. Sat in the truck for another minute. The engine ticked cooling. I pulled the key from the ignition and set it on the passenger seat with the chain-of-custody copy I'd designated for the truck file. Tucked that copy into the glove box on top of the field manual. Locked the truck from outside with the manual lock because I was leaving the key fob inside, and walked the key and fob up to the front desk in a hotel envelope I'd taken from the room.

The desk clerk asked who the keys were for. I told her they were for a state EPD vehicle in the parking structure, level three, and that a man named Brandon Wheeler from the Brunswick field office would be sending someone to collect them. I wrote Brandon's office number on the outside of the envelope in pen the clerk handed me. She took the envelope and put it in a numbered slot behind the desk without asking further questions.

Back in the room, I sent Brandon the photo log and the typed documentation file as email attachments. Then I sat down on the edge of the bed again and looked at the river.

Brandon hadn't called back. He might have been on the phone with someone I'd never heard of, or driving north himself, or standing at Station Eleven watching the bridge with Armstrong. I didn't know. I wasn't going to call again.

The work was in the right hands now — the samples in transit to Atlanta, the documentation on record, the truck available for retrieval. Whatever the agency did next, they had everything I'd collected. I'd followed every protocol that applied and a few that didn't quite fit, and the paperwork was cleaner than most contamination events I'd processed.

The entity had crossed the dry creek bed at Station Ten and kept moving. It had been past Station Eleven before Armstrong arrived. It had been past Station Seven before I got there at dawn. Whatever it was heading toward, it had covered that distance while I was still pulling samples and logging coordinates and trying to fit what I was seeing into categories built for something else entirely.

It was already there. Had probably been there for hours.

I watched the river run east.