Exam papers are counted in sealed packets, signed for, counted again into the hall and counted out of it, and every script carries a candidate number that maps to a living person on a list. One a year does not.
I have run the exams at a comprehensive for fourteen years and I have never had a breach.
The counting, because the counting is the whole of this.
Question papers arrive in sealed packets with a printed count on them. They go into a safe in a room with two locks and an alarm, and the safe is opened on the morning of the exam in front of two people who sign a log.
In the hall, papers go out one per desk, and the desks are on a plan I drew. Each candidate writes their candidate number and their centre number on the front of the answer booklet. The number is four digits and it comes off the list the board sends us, and every number on that list is a person whose birth certificate I have seen a copy of.
At the end you collect in rows, you count, you band them, and the count has to equal the number of candidates who were in the room. Absences are recorded to the minute.
Then the scripts are sealed into a bag with a tamper evident strip and a dispatch note with the number written on it, and a courier takes them.
So there are four counts: packets in, papers out, scripts in, scripts dispatched. They have to agree.
Once a summer series, in one paper, there is one script too many.
Fourteen years. Eleven of them have had one. Never two in the same series, never in the same subject twice running.
The script has a candidate number on it.
Four digits, in the right range for this centre, written in the box, in pen, in a hand that is different every time.
It is never a number on the list. It is never a number that has been used by anybody at this school in any year I can check, and I can check back to 2003 because the old registers are in the same cupboard.
The centre number is ours and it is correct.
What is in it.
The paper is attempted. Not completed, and not nonsense.
I am allowed to look at a script in exactly one circumstance, which is reconciliation, and that is what this is, so I look at the front and the first page and no further, and I have kept to that.
The handwriting is a candidate's handwriting. The questions attempted are the questions on that paper that year. The work is ordinary. In the 2019 maths paper the first three answers were right.
What has been done about it.
Every one has been reported. The procedure for an unidentified script is to report it to the board, which I have done eleven times, with the paper, the date, the number written on it and a statement of the counts.
The board's process is that they investigate, they check the number against their own records, and they write back. All eleven letters say the same thing in slightly different words: no candidate of that number is registered at this centre, the script cannot be marked, retain it.
So they are in a box in the locked cupboard. Eleven scripts, labelled, with the correspondence.
Nobody has ever suggested I am miscounting. That is the thing that gets me, because a miscount is the obvious answer and I would take it. A miscount does not produce a piece of paper with writing on it.
What I checked in 2021, which is why I now count the desks myself twice.
The paper.
Exam stationery is printed with the series and the board's mark on it, and the booklets are sequential. Not numbered where you would see it, but there is a code in the footer and it runs in order through a packet.
I pulled the codes off all the scripts from one paper in 2021 and put them in order.
The packet ran in sequence, every booklet accounted for, first to last, with no gap.
The extra script's footer code is in the sequence too. It sits between two booklets used by two real candidates who sat four desks apart, which is not where the collection order would have put it, and which means the booklet came out of our sealed packet.
We opened that packet in front of two people. I was one of them. The printed count on the seal was right.
What I do now.
I count the blank booklets into the hall, and I count them out, and the two counts agree, every time, which I want on the record because it means I am not handing out a spare.
I have stopped invigilating the back rows. I sit at the front on a chair where I can see the whole room, which is what you are supposed to do anyway.
And in June, before the first paper, I write the number of candidates on the board outside the hall in marker, the way I always have for the caretakers, and then I write it again in my own book, and in the last three years I have gone back and checked my own book against the list between papers, because the one thing I have not told the board is that twice, in 2022 and last year, the number in my own handwriting was one higher than the list, and I do not remember writing it.