Tuesday, the sixteenth of May 2051, twenty to three in the morning, Ken Schaefer's ditch.
Every dot on her C-Q plot has an F written next to it, because in two years Janelle has never once beaten a storm to this pipe. She drives out after the rain like everybody drives out after the rain, and what she bottles is the back half: the water already on its way down, already thinned by everything that fell on top of it. Falling limb. She put that in the title of the plot in capital letters so nobody would forget, herself included.
So last night she set the alarm off the radar instead of the clock. The county weather agent pushes a cell track to her phone, and she gave it one job: wake me when the line crosses Pekin. It crossed at 1:50. She was in her boots at 1:52, and the kid, half asleep on the couch where he'd lost an argument about bedtime, said "Is it the pipe," and she said "It's the pipe," and he said "Bring me a doughnut," and went back under.
Headlamp. Rain loud enough on the hood that she can't hear the pipe until she's standing over it. The V-notch is only just waking up, a thin tongue of water over the point of the plate, and the staff reads 0.21 and climbing while she watches. She stands in the ditch with the twenty-liter bucket and the meter and does the one thing her job is mostly made of, which is waiting for a number. She wants 0.30. Not the peak. A head she can count on seeing again on the way back down.
3:20, 0.30, rising. Bucket 17.9, 18.2, 18.0 seconds. Call it 1.1 liters a second, right on the curve, which is her curve, fit to her own buckets, because the plate stopped acting like a handbook V the first summer and the buckets are the reference. EC 700. pH 7.16. The water is browner than anything she has bottled here, first flush, whatever's been sitting in the tile since the last rain finally getting pushed out. Duplicate, split, before bottle, capped underwater. She writes R on every cap with the grease pencil, pressing hard, so a cold lab tech at ten in the morning cannot misread it.
The peak comes at 4:15, 0.38. She doesn't sample it. A peak happens once per storm, so it can't be compared with anything, and she's down to bottles she wouldn't trust anyway. She sits in the truck with the heater on her wet knees and eats the doughnut she was saving for Dale, then the one for the kid, and decides Dale can have the next storm's.
5:40, 0.30 again, falling. Bucket 18.1, 17.8, 18.0. Same flow, as near as a bucket can tell. EC 610. pH 7.21. Clearer water. F on the caps.
Same flow, two waters. Ninety microsiemens apart at the same liter a second, and the only thing that changed was which side of the storm she was standing on. That's the size of the thing her plot has been ignoring, and her plot had the decency to say so in its title.
Dale runs the before bottles at ten on the surplus Hach. Rising 3.0. Falling 3.2. He holds up a cap and squints at the R.
"You were out there at three."
"The storm was out there at three."
It's after he leaves that she sees it. If falling water were just rising water with rain stirred in, the alkalinity would fall with the conductivity, because rain carries next to none. It went up. Conductivity down ninety, alkalinity up two tenths. The two halves aren't one water at two strengths. They're two waters: whatever the first flush scoured out of the tile on the way up, and something slower and older, more like groundwater, on the way down. She writes NOT DILUTION ONLY on the card and puts a box around it.
"Wendell paying you for that?"
"Gray hours. Ken didn't ask for it."
The storm counts once on the tally, not twice. Two bottles don't make two storms, and twelve before-storms is twelve storms, not twelve bottles. On the back of card thirteen, a fourth pencil stroke.
223. 5/16/51 S21 0.9 IN. FIRST RISING-LIMB GRAB. H 0.30 R @0320 EC 700 pH 7.16 / H 0.30 F @0540 EC 610 pH 7.21. BUCKET 1.1 L/S BOTH. SAME Q, DIFF WATER (CLOCKWISE). ALK R 3.0 F 3.2 (DALE): EC DOWN, ALK UP = NOT DILUTION ONLY. CURVE = BUCKET FIT. PEAK H 0.38 @0415 NOT SAMPLED. TALLY 4/12. 4.1 H, GRAY.
Tuesday, the twenty-third of May, quarter past nine, Wendell's office.
Quarterly budget review. Wendell prints the line items because, he says, a screen lets you skim, and he doesn't trust anything he can skim. The district's budgeting agent has already flagged three lines in yellow as "variance-likely." Wendell has ignored all three and circled a fourth in red pen.
LAB - ALK, TAZ-17, PRE-APPLICATION SERIES. $60/SAMPLE.
"A before of what?"
"Of Ken's lime. He'll spread the south half when the soil test says. Probably fall."
"And then."
"Then somebody asks whether it did anything to the water. Terraline asks, or Ken, or IEPA. The answer is after minus before. If there's no before, a model makes one up."
"They already made one up."
"Out of one storm of mine. I'd like the next one made out of twelve."
He taps the red circle. "Twelve at sixty is seven-twenty. Plus Dale's after-hours when you come in with bottles at six."
"Dale doesn't charge after-hours. He charges doughnuts."
"Doughnuts aren't a line item."
Ray, passing the door with a coffee, not stopping: "They should be."
She puts the second sheet on the desk. Marcy Kuhl's creek forty, second point, which Dale ran Friday: 3.0. Ken's spring storms, 3.1 to 3.4. Marcy's field never got lime, never got ash, never got anything except a satellite agent trying to sell her some, and it reads within a few tenths of Ken's.
"That's the paired field. It's the only reason Ken's before means anything. If Marcy's water moves and Ken's moves the same, it's the year. If only Ken's moves, it might be the lime."
Wendell reads both pages. He holds his glasses by one arm, the way he does when he's deciding whether he's being had. "Twelve. Not thirteen. If you need thirteen you come back and tell me why."
"And Marcy's?"
"Marcy's is in the twelve."
"It isn't. It's a different pipe."
A long pause, the kind where the budgeting agent would have filled the silence with a suggestion if Wendell let it talk. He writes "+6 KUHL" beside the red circle and initials it.
"Six. Monthly. Don't make me learn what a limb is."
She winces at monthly and takes it. A storm series on Ken's tile against a calendar series on Marcy's is two different kinds of water, and a control only controls if it's sampled the way the treatment is. But six monthly bottles are six more than nothing, and she can at least say on every one what the creek was doing. On the Kuhl sheet, that night, she adds a column called FLOW_TYPE and fills in the first two rows.
224. 5/23/51 BUDGET REV W/ WENDELL. ALK PRE-APP SERIES APPROVED: 12 @ $60 TAZ-17 + 6 KUHL CREEK 40 (MONTHLY). KUHL 2ND PT ALK 3.0 (DALE 5/19) VS TAZ-17 SPRING 3.1-3.4. KUHL = CALENDAR, NOT STORM: FLOW_TYPE COL ON KUHL SHEET. "IF YOU NEED THIRTEEN COME BACK AND TELL ME WHY." 0.6 H, GRAY.
Tuesday, the thirtieth of May, five past seven in the morning, her desk.
The methods PDF lands in the district's partner inbox at 4:11, which is when Terraline's agent does its mail. Methodology v2.3, effective Q3 2051. Prospective, like they promised. Nothing already paid gets touched. Vintage terms.
Page 6:
Sulfate concentration trends at reference sites are now normalized for discharge using a dilution correction derived from District concentration-discharge analysis (TAZ-17, 2051-04-25). Correction factor 0.90 applied across all enrolled sites.
She reads it twice. They took her plot. That part is fine, it's a public record, and the plot is right that most of her falling pulse was rain. They believed her about the rain. It's the first time anything she posted has made Terraline's model more honest instead of just more confident, and she would like to feel good about that for longer than she does.
Then she opens their Figure 3. Her five dots, redrawn in Terraline blue, on a nice clean slope. No open circle for S16, the storm with the flagged head that she plotted hollow so nobody would lean on it. No F next to anything. Her title said C VS Q, 5 RATED STORMS, FALLING LIMB ONLY, NOT A TREND. Theirs says District C-Q relationship, TAZ-17.
The label didn't survive the redraw. Titles don't. A title is the first thing a re-plotter types over, whether it's a person or an agent, and the redraw came out of the 4:11 mailbox, so this one was an agent.
And 0.90 is her ten percent residual turned into a constant and pointed at fourteen fields she has never sampled. Different soil. Different tile, some of it older than Ruebush's. One pipe's falling half, five points, becoming everybody's rain.
She doesn't know yet whether the factor is too big or too small, and it's the wrong question. Their 0.90 is a dilution correction. Two weeks ago her rising bottle read 700 against a falling 610 at the same flow, with alkalinity going the other way, which isn't dilution, it's a different water. The size doesn't matter if the mechanism is wrong. Their correction is fit entirely from the half of the storm she happens to see when she drives out after it, and it has no information about the other half at all.
Ray reads it over her shoulder, coffee steaming on her keyboard. "Congratulations. You're a methodology."
"I'm a figure with the caption cut off."
She prints page 6 and staples it behind her original plot in the folder, so the two sit together. The one that says what it is, and the copy that doesn't.
225. 5/30/51 TERRALINE METHOD v2.3 (EFF Q3 2051): DILUTION CORR 0.90 FLEET-WIDE "DERIVED FROM" TAZ-17 C-Q 4/25/51. FIG 3 REDRAWN: NO S16 OPEN CIRCLE, NO LIMB LABEL, TITLE DROPPED. 1 PIPE, 5 PTS, FALLING ONLY -> 14 FIELDS. FACTOR FIT ON F ONLY; R NOT IN SAMPLE. S21 EC DOWN/ALK UP: NOT DILUTION ONLY. 0.9 H, GRAY.
Tuesday, the sixth of June, half past ten, her desk.
The obvious move is to write to Terraline and tell them their factor leans on half a storm. She drafts it in her head on the drive in, past the elevator and the solar lease signs, and deletes it before she reaches the parking lot. Every sentence she hands their agent comes back in a methods PDF under somebody else's logo, and then she's designing their product for free. The district doesn't design products. It keeps records.
So she fixes the record.
The public C-Q file used to be a table with a title. Now it's a table with columns, and the caveat lives in the columns: STORM, DATE, TIME, H, Q_LPS, Q_METHOD, LIMB, EC, PH, ALK, HEAD_FLAG, USE. Every row says F or R under LIMB. S16 says FLAGGED_HEAD and USE: NO. S17 goes in too, with ICE under HEAD_FLAG and NO under USE, because a storm left out is a storm somebody else gets to fill in. S21 is two rows, R and F, same Q, ninety microsiemens apart, one above the other, so the disagreement is in the file instead of in her head.
Anybody who redraws it now has to delete a column or carry it. Deleting a column is a decision, and a decision has a name on it. Losing a title is just Tuesday.
The same morning the partner inbox has a polite request from the Terraline agent:
Would the District consider providing TAZ-17 storm observations via a machine-readable endpoint to support timely methodology refinement? A partner API key has been pre-issued for your convenience (valid 12 months).
The key is attached, a little file, already hers if she wants it.
She reads the request to Wendell over the cubicle wall. "They want a pipe from our pipe."
"Do we have to?"
"It's all public already. They can download it like anybody."
"Then what's the difference?"
"A feed is a promise to be on time. Then a late storm is a gap, and they fill gaps."
Wendell thinks about that longer than she expected. "Say no nicely."
She replies in one line:
District storm data remain available as published files. The District does not operate a partner feed.
Then she deletes the key from the attachment folder without opening it, and adds a row to the TAZ-17 site history, first field dated so the parsers read the date before anything else: 2051-06-06, THIRD-PARTY REQUEST, MACHINE FEED, DECLINED.
226. 6/6/51 PUBLIC C-Q FILE REBUILT: CAVEATS AS COLUMNS (LIMB, HEAD_FLAG, USE). S16 FLAGGED/NO, S17 ICE/NO ADDED, S21 2 ROWS R/F. NO LETTER TO TERRALINE RE FACTOR (NOT OUR DESIGN). TERRALINE REQ MACHINE FEED + PRE-ISSUED KEY: DECLINED (WENDELL OK), KEY NOT OPENED. 1.4 H, GRAY.
Tuesday, the thirteenth of June, ten to six in the morning, Ken's ditch.
An inch and six tenths overnight, the kind of June rain that comes in sideways and leaves the beans looking surprised. She set the radar alarm again and slept straight through it, which is how she learns the alarm was set to vibrate and her phone was face-down on a library book. So: falling limb, like always. She writes F on the caps before she uncaps them, so she can't forget to be honest.
Staff 0.41. Twenty-liter bucket 9.9, 10.1, 9.8. About 2.0 liters a second; the curve says 1.95. EC 560, pH 7.24, the lowest conductivity this pipe has given her since the trench went in. Big storm, weak water, right where the slope says it should land. She holds the meter a moment longer anyway, partly out of habit and partly because a reading that lands exactly on the line is the only thing in this job that ever agrees with her.
Duplicate, split, before bottle. Dale, at ten: 2.9.
Ken's truck comes down the field road at seven. The beans are up in the rows behind him, two leaves each, very green against the mud. He stops beside the ditch and doesn't get out, which is how Ken has every conversation he considers optional.
"How many."
"Five."
"Out of twelve."
"Out of twelve."
He nods at the V-notch, still pouring over the plate. "Pulled the soil samples Saturday. South half. Gave them to the co-op Monday."
She knew it was coming. It still tightens something under her ribs, the way the first fizz of acid on a core does when she hasn't decided yet what she wants it to say.
"When do you get the numbers?"
"Two weeks. Then I'll know if I'm liming. Then you'll know." He looks at her clipboard, the tally side up. "Seven more storms between now and October, you've got your twelve?"
"If it rains seven more times. July doesn't always."
"Well." He puts the truck in gear. "I can't help you with that."
It's true, and it's also the closest Ken Schaefer has ever come to wishing her luck.
She stands in the ditch after he's gone and does the arithmetic she's been avoiding. Five befores in hand. Seven to go. Maybe four months, if he limes in October, fewer if the co-op's agent tells him August. Every one of those storms will have to come at night or come big or come with a head she can trust twice, and she can't schedule any of that. The only thing she can schedule is the alarm, and she can at least turn the sound on.
On the back of card thirteen, a fifth stroke. Under the tally, smaller: SOIL TEST AT CO-OP 6/12. RESULT ~6/26.
227. 6/13/51 S22 1.6 IN. FALLING (ALARM ON VIBRATE). H 0.41, 20 L BUCKET 2.0 L/S VS CURVE 1.95. EC 560 pH 7.24 (LOWEST SINCE TRENCH, ON SLOPE). ALK 2.9 (DALE). TALLY 5/12. K.S. PULLED S-HALF SOIL SAMPLES 6/10, TO CO-OP 6/12, RESULT ~2 WKS. 2.8 H, BLACK (K.S.).