Noeline Farrant wrote down the two hundred and fourteen thousand, eight hundred and sixth line of her working life on the fourteenth of January, at twenty past eight in the morning, in a book with a red cloth spine that has to be turned on its side before it will open properly. The line read: twenty six, oh four one seven. Aroha Bell. Fifty eight. Ward three. Requesting doctor Hollis. Yellow top. Drawn ten forty. Bench three five oh five. H. She works in one room at Palmerston North Hospital, with a hatch, a bench, a printer that eats labels, a hole in the window screen that somebody has covered with a page of a telephone directory, and two hundred and forty-six metres of shelving holding the completed books. The books go to the archive in grey boxes and the boxes are kept in a storey nobody visits. The register goes back to 1994 in a run of identical red volumes. The run before that is on microfilm in the same room, in a cardboard box with the lid taped, in the cabinet by the door.

There is a screen on the desk that could have done all of it. Every specimen in the building gets a barcode, and the barcode goes into the laboratory information system, and the system knows the accession number and the patient and the request and the result and the time the tube came off the van and the time the analyser finished and who looked at the result and when they looked at it and whether they printed it. The system is very good. It cannot tell you when the tube got warm. That is the whole of Noeline's argument. She has never had to make it out loud. A tube is an object with four facts attached to it, and the four facts decide what the numbers will be: how old it is, how warm it is, what is in the lid, and how much of it there is. The electronic record knows about the third of those, because the analyser is fed the tube and forgets it within the second. It knows nothing about the first. It knows a great deal about the last.

Gwen Iverson taught her that, in 1988, standing at the same bench with a cigarette she was not allowed to light. Write down what the tube was before you write down what it said. Gwen held the pen for twenty-two years and she retired in 2009 to Feilding and she rings Noeline on a Sunday about twice a year, and both of them still say the sentence.

Aroha Bell is fifty-eight and drives a school bus on the Pahiatua Road, which is forty minutes each way with forty-one children on it. She has been tired since before Christmas in a way that has not improved, and on the twelfth of January she said, in the consulting room, that her hands go dead halfway through the afternoon. Not numb. Dead. She said it the way you say a thing you have said out loud in the car and to yourself and to nobody else. Her general practitioner is Dr Hollis, and she ordered a full blood count, an electrolyte panel and a renal panel, and a practice nurse took the blood at ten forty on the fourteenth in a room with a strip curtain and a window onto the car park. The courier was due at 11:15.

The courier was due at 11:15 until the sixth of January, when the route out of Feilding was merged with the route out of Levin to save a wage, and the practice lost the eleven fifteen collection and the next one is at ten past three. This is not a decision anybody made about Aroha Bell. It is a document. It has a date on it and it was signed by somebody whose job it was to save a wage. So the tube stood on a bench in Feilding for four hours and twenty-five minutes, in a box, and then went into a van in the Manawatū in the middle of January with no cold box in it, because nobody had ever asked for one, because the eleven fifteen van was always twenty minutes.

Four hours and twenty-five minutes is in the register because the practice telephoned the draw time, and it is in the register and nowhere else. The barcode on the lid was scanned off the van and it knows the arrival and nothing before the arrival. The only written record on earth of the age of that specimen at the moment it was analysed is the line Noeline wrote at twenty past eight in the morning of the fourteenth of January, in a book that has to be turned on its side.

The panel came off the analyser at 15:31 and the potassium was 7.1 millimoles a litre. The analyser also printed a flag, in the column the analyser owns, one letter long, reading H. H means the red cells in that tube have burst, and the potassium that was inside them is now in the plasma being counted. Seven point one is above the telephone threshold. It was telephoned at 15:58 to the practice, and Aroha Bell was told to drive to Palmerston North, and she drove there herself and was sitting in the emergency department at twenty past six with a twelve lead electrocardiogram on the wall and a cannula in her hand.

What happens in an emergency department in New Zealand to a person whose potassium reads 7.1 is well defined and it happened to her. Ten millilitres of ten percent calcium gluconate, given slowly, which does nothing whatever to the potassium. It does not lower it and it is not supposed to. It buys the heart muscle's membrane against the potassium, so that the potassium does not open the fast channels and stop the heart in a rhythm that will not pump. It is a race run against time with a prize that is somebody else's job. Then something is given to get the number down. Insulin and glucose, so the cells take the potassium in through their doors and sit on it, which is measured in hours and wears off. Salbutamol through a nebuliser. Somebody in the building now gave an injection and an aerosol to a woman with normal kidneys, because of a tube.

The repeat was drawn at 21:10 in the department, forty minutes after the first bag went up, and it ran on the analyser eleven minutes after it was drawn, because the emergency department specimens do not wait for a van, and the potassium came back at 4.4. After that it is settled. The 7.1 was not her. It belonged to a tube.

What Noeline could not put down was the arithmetic. Her training says a potassium in whole blood left standing at room temperature rises by about a tenth of a millimole a litre an hour. The cells do not stop. Potassium leaks out of them and glucose goes the other way, and a specimen that has been in a warm van all morning is not the specimen that left the arm. Four hours and twenty-five minutes at a tenth an hour is four tenths of a millimole. She put a woman on a trolley at twenty past six in the evening over four tenths of a millimole.

So the delay was real and it was not enough, and the only other thing in that tube was the flag she had copied onto her own line, and the flag is not a guess. The flag is an index. The analyser shines light through the specimen and measures how much red light gets through, and the index runs from zero to four plus, and it is roughly calibrated so that a plasma with five grams of haemoglobin in a litre reads about one plus. A one plus is a nuisance. A three plus is a specimen nobody should have believed, and a three plus carries its own potassium with it at a rate the policy sheet on the wall states as between one and two millimoles a litre for every ten grams. Twenty-six oh four one seven is flagged three plus.

Four tenths of a millimole of delay and up to three of haemolysis will make a number in the low sevenes out of a woman whose real value is four point four, and the laboratory's rule for a three plus is not to footnote the result. The rule is that the specimen is not reported and a fresh one is asked for. That rule exists on paper. It exists on the screen as a greyed field at the bottom of the report with two words in it that nobody reads: values unreliable. The flag was in the record from 15:31 and the result was telephoned at 15:58 and the telephone was not supposed to pass on a three plus without a fresh specimen. The telephone call is not a note in this hospital's system. It is a record with a name and a time and four boxes, and the boxes are: the number read out, and the quality of the specimen read out. The call at 15:58 has the number in the first box and the second box is empty.

The name in the record is R. Fenwick, who was rostered that afternoon, who is twenty-six, who is on the fourth year of a degree, who did the call in the correct order of the form, which is the value and then the specimen, and who left the second box empty because nobody had ever shown him that anybody reads the second box.

Here is what Aroha Bell has, and it is not a story anybody will be told at the bedside. She has two nights in Ward three. She has four days of an injected drug and an aerosol she did not need. She has a discharge summary that says hyperkalaemia, cause not established, which is the phrase that keeps a thing open forever. She has a losartan she had been on at a hundred milligrams a day for two years, stopped on admission by a registrar who was reading a potassium of 7.1, and not restarted on discharge, because the reason it was stopped was a reason that never existed. Her blood pressure has been 154 over 92 every reading since the twenty-second of January. Nobody has changed it back because there is nothing in the record to change back. The record says the drug was stopped for a reason. And on the twelfth of November, eleven weeks before any of this, the same practice drew a full blood count from Aroha Bell at 09:20, and the courier that day was the old route, and it collected at 16:10, six hours and fifty minutes in a van.

That specimen was also flagged three plus. The haemolysed film could not be reported accurately and the report said so in a sentence a third of the way down the page. The haemoglobin was 76 grams a litre. The mean cell volume was 106 femtolitres, which is outside the range for an adult by six, and it was reported, and it was commented on, and it was not telephoned, because a haemoglobin of 76 and a red cell size of 106 are the kind of number that goes in a letter. They have been in three letters. Aroha Bell has three letters on her file that say her red cells are small and she is anaemic and nobody has ever rung about the size.

And her hands are not the potassium. A potassium of 7.1 gives heaviness in the limbs and a weakness in the muscles and a heart that may misbehave in a particular way. It does not give a school bus driver dead fingers at three in the afternoon in February. There are a small number of reasons for dead fingers in the afternoon, and one of them is that your red cells are the wrong shape to carry what they carry, and one of the anaemias that does that has been sitting in her file since November behind a flag that does not shout unless the potassium is loud. A referral went on the twenty-ninth of January. It has a date written in the top right corner and the date is the fourth of April.

She took it to Dr Ines Varga, who is fifty-one and the consultant chemical pathologist and responsible for the telephone threshold and for the false positives, and who is not a fool and does not enjoy this. Varga's position is that the flag is not a failure, it is an honest measurement, and the report carries it, and the person at the end of the report has to look at it. One chemistry specimen in thirty comes back with the flag set, and most of them are fine, and if the flag stopped everything then she would spend her life on the telephone about nothing. Noeline's position is that nobody looks at anything on a report they expect to be boring, and that a flag which can be read past is not a control, it is a decoration, and the hospital had known which of the two it was since the day the analyser was installed, and had decided to call it a comment so that nobody would have to say so out loud. They agreed on one thing, which is that it should be measured before it was argued about. Noeline had nine weeks and a spare ledger and counted every chemistry specimen that came off that analyser in a working day.

The count is worth setting down because it decides the argument and not the way either of them expected. On the seventeenth of February, in one working day, sixty two chemistry specimens came off that analyser. Forty one of them had been in the building less than two hours. Twenty one had been in it more than three hours. Of the twenty one, nine were flagged three plus or worse. Of the forty one fast ones, two were. The delay was doing almost nothing. Half the flagged specimens had arrived within the hour. The van was not the story. What the flagged ones had in common was on the register, in the column Noeline writes for the practice and not for the tube. Nine of the eleven were first attempts. Seven were drawn by the same practice nurse who does Thursdays in Feilding. Five were drawn with a tourniquet that was up for more than a minute. Four were from hands that were cold, and two were from a man who had come in from a shed and whose hands were the wrong shape for it.

So the signal was not the van. The signal was a person on a bad day, meeting a small vein, and getting one go at it, and the red cells bursting on the way into the tube because the needle was small and the hand was cold and the first go was the only go. Noeline read that column for four hours on the Saturday and then went and sat in her car in the staff car park, which is not a thing she does. The change went in on the ninth of June and it took two lines of code in the middleware that sits between the analysers and the record. The comment line now prints the elapsed time with the flag. It reads, in the corner of a report, for a specimen like Aroha Bell's, H, RECEIVED 4H25 FROM DRAW.

The haematology side liked it. The chemistry side complained about the width of the comment field for a month. The telephone calls went up by four a week for a fortnight and then came back down, and the four a week that stayed were mostly a haemoglobin of 84 and a creatinine of 96, mild things that under the old arrangement would have sat in a letter until the next appointment and would now get a phone call instead, which is four people a week finding out something about themselves on a Tuesday rather than in March. Noeline was right that the tube was the culprit. She was wrong about the delay, and she had been wrong about it for five months, and the system that was built to answer her turned out to be answering a different question, and nobody in the building has ever told her that the two arguments are not the same argument.

Aroha Bell came back on the third of April, at 09:50, because the appointment before it had been moved by somebody who did not know why it mattered. The courier took the tube at 11:02, and the bench had it at 11:31, and the analyser finished at 11:38, and the potassium was 4.6 millimoles a litre and the haemolysis index was 1 plus, which is the flag for a trace, and one plus prints too. Noeline wrote the line and put the book back on the shelf in its alphabetical place with the other two hundred and forty-five books. The line was correct in every part of it. Aroha Bell walks out with a number that is four point six, which is not four point four and not 7.1, and a red cell volume of 105 that is still six points too large and is on a list, and an appointment on the fourth of April that is tomorrow and is a waiting list with a date on the top corner of it. Noeline Farrant retires on the thirtieth of September. The position is not advertised. There is a form for it on the wall by the door and the form has been on the wall since 1994 and the last person who filled it in filled it in with the wrong date.