Guide · England and IB DP

How AQA, Edexcel, OCR and the IB mark an error carried forward

Every board lets a student keep the marks for correct work on a wrong earlier answer. They differ on the working they need to see and on where the allowance stops.

The Graidable team

Published 8 October 2026

Question 1: a straight line passes through $A(1, 3)$ and $B(5, 11)$ . Find (a) its gradient [2] and (b) its equation in the form $y = mx + c$ [2].

Student’s answer, Question 1

Marks, with ft

\text{(a)}\;\; m = \dfrac{11 - 3}{5 - 1}

m = \dfrac{8}{4} = \htmlClass{lab-err}{3}

A08 ÷ 4 = 2

(a) 1 [2]

\text{(b)}\;\; 11 = 3 \times 5 + c

M1uses own m

c = -4

y = 3x - 4

A1ft

(b) 2 [2]

Marking rule	(a) [2]	(b) [2]	Total [4]
With follow-through	1	2	3
Without	1	1	2

Figure 1. Constructed example. A slip in part (a), marked with follow-through: the error costs one mark, not two.

In this article

One error costs marks once
The four boards use different names for the same rule
Follow-through needs working on the page
Where follow-through stops
Write the rule into your own mark scheme

A student gets part (a) wrong, then uses that wrong answer correctly in part (b). Error carried forward (ECF) is the rule that part (b) still earns its marks. Maths mark schemes call the same thing follow-through and write it “ft” or “FT”.

AQA, Edexcel, OCR and the IB all apply it, and each publishes how. Their rules agree on the principle and differ in the details that decide real marks: whether the working has to be shown, which kinds of error qualify, and what happens when the wrong value is impossible. Those details matter when you mark a mock with a board’s mark scheme, and when you write a mark scheme for your own test.

One error costs marks once

Here is a two-part question and its mark scheme.

Question 1 A straight line passes through $A(1, 3)$ and $B(5, 11)$ .

(a) Find the gradient of the line. [2]

(b) Find the equation of the line in the form $y = mx + c$ . [2]

Table 1. Mark scheme for Question 1
Part	Mark	What earns it
(a)	M1	$(11 - 3) / (5 - 1)$ , or an equivalent calculation
(a)	A1	2
(b)	M1	A point on the line and their gradient substituted into $y = mx + c$
(b)	A1ft	$y = 2x + 1$ , or the correct equation for their gradient and the point they used
Total	4

A student who sets up the gradient correctly and then writes $8 / 4 = 3$ loses the accuracy mark in (a). In (b) they do everything right with their 3. With follow-through they keep both marks there, and without it they lose one of them for a mistake already paid for in (a) (Figure 1).

AQA’s marking guidance for GCSE Mathematics gives the reason in one line: the ft mark is there so that a student isn’t penalised twice, once for the earlier step and again for the final answer¹.

The mark scheme above has a detail that is easy to miss. With the wrong gradient, the points $A$ and $B$ no longer give the same line: a student who substitutes $A$ gets $y = 3x$ , and one who substitutes $B$ gets $y = 3x - 4$ . Both have followed through correctly, so the A1ft line has to accept either.

The four boards use different names for the same rule

Table 2. How each board writes and limits follow-through
Board and paper	Written as	What the board’s guidance says
AQA GCSE Mathematics	ft, A1ft	Method marks always follow a wrong value that came from a correct calculation. The final accuracy mark follows only where the scheme says A1ft¹.
AQA A-level Physics	ECF, conseq	Likely to be limited to calculation questions, and shown in the scheme. An arithmetic error costs one mark only².
Edexcel GCSE Mathematics	ft	Applies where the question’s own scheme allows it. A one-step follow-through can be credited without working; with more steps the working must be seen³.
OCR A Level Physics A	ECF	For numerical answers unless the scheme says otherwise. Given after arithmetic, transcription and power-of-ten errors, not after a wrong equation⁴.
IB Mathematics: analysis and approaches	FT	Applies between parts of a question, and working must normally be present. Inside a part, an error ends the accuracy marks but not the method marks⁵.

Mark schemes signal follow-through in the wording of each line as well. The word “their”, as in “18 − their 12”, tells you to accept the student’s own value. AQA and the IB both use it that way¹^,5.

Edexcel adds two finer signals. A number in inverted commas, such as “12” × 50, must come from a correct method, though the arithmetic may have gone wrong. A word in square brackets, such as [area] × 1.5, accepts whatever value the student believes is the area, however they reached it³.

Follow-through needs working on the page

A wrong answer on its own is hard to follow through, because you can’t see that the student used their earlier value correctly. The boards handle this in slightly different ways.

Edexcel’s rule depends on the number of steps. If a follow-through answer takes a single calculation, you can check it yourself and award it without working. If it takes more than one stage, the working has to be there, even when only one route could have produced the answer³.

The IB normally wants working for FT marks. The exception is a later part where every mark is for the answer or is implied: there the student’s correct follow-through answer earns the marks without working⁵.

AQA limits how far follow-through goes inside one part. A student who slips cannot reach full marks on that part through an A1ft. Full marks stay possible only where the question is split, so that a slip in (a) leaves (b) open¹. The IB says the same thing in mark terms: within a part, once an error is made, later A marks that use it are gone, and M marks can still be awarded⁵.

Where follow-through stops

Follow-through rewards correct work on a wrong value. It was never meant to rescue wrong work, and each board says where the line is.

A wrong method or equation. OCR’s physics mark schemes separate kinds of error. After an arithmetic error, a transcription error or a power-of-ten error, the marker withholds that mark and gives ECF on what follows. A wrong equation counts as “wrong physics” even if it happens to give the right number, and it gets no ECF within the question⁴.

An impossible value. In the IB’s instructions, if the error leads to an inappropriate value, such as a probability greater than 1 or a non-integer where an integer is required, the final answer mark is not awarded⁵. The method mark is. Figure 2 shows a student whose slip gives a probability of ${20}/{15}$ and then a negative one.

Question 2 A bag holds 5 red counters and 3 blue counters. Two counters are taken at random without replacement.

(a) Find the probability that both are red. [2]

(b) Find the probability that at least one is blue. [2]

Question 2: two counters are taken without replacement from 5 red and 3 blue. Find the probability that (a) both are red [2] and (b) at least one is blue [2].

Student’s answer, Question 2

Marks

\text{(a)}\;\; \dfrac{5}{8} \times \dfrac{4}{7}

= \htmlClass{lab-err}{\dfrac{20}{15}}

A08 × 7 = 56

(a) 1 [2]

\text{(b)}\;\; 1 - \dfrac{20}{15}

M1uses own value

= -\dfrac{5}{15}

A0not a probability

(b) 1 [2]

Figure 2. Constructed example. The method in part (b) follows through, but the answer cannot be a probability, so the accuracy mark is withheld.

An error that makes the question easier. If a wrong value turns a quadratic into a linear equation, the student has done less than the marks were written for. The IB tells examiners to award fewer FT marks in that case, by thinking about what each mark is for⁵.

A “show that” answer. The value is printed in the question, so there is nothing to carry forward. Under the IB’s rules, a student who fails the “show that” and then uses their own wrong answer in later parts, in place of the given value, gets no FT marks there⁵.

A note in the scheme. Individual questions can switch follow-through off. OCR’s June 2018 Modelling physics scheme has a line reading “No ECF from (a)(i)” against one mark⁴. The note on the question always wins over the general rule, and Edexcel says so at the top of its guidance³.

A misread sits next to all this. A student who copies 83 for 38 from the question has made a slip of the same kind, and the boards treat it in the same spirit. AQA awards the method marks as usual for a genuine miscopy, and not the A or B marks. Edexcel allows method marks provided the misread hasn’t simplified the question. The IB withholds the first mark and awards the others¹^,3^,5.

Put together, the limits come down to four questions you can ask about any carried error (Figure 3).

Is the method in the later part itself wrong?
Yes: No follow-through. There is no correct work to credit.
No: go on to the next question
Was the correct value printed in the question, as in a “show that”?
Yes: No follow-through. The student should have used the given value.
No: go on to the next question
Is the answer impossible, such as a probability above 1?
Yes: Give the method marks and withhold the final answer mark.
No: go on to the next question
Did the error make the later part much easier?
Yes: Give fewer marks: only for the steps the student still had to do.
No: go to the last step
Otherwise: Follow through. Mark the later work as correct for the student’s own value.

Figure 3. Four checks before a carried error earns follow-through marks, drawn from the boards’ published guidance.

Write the rule into your own mark scheme

A board’s mark scheme arrives with these decisions made. Your own test doesn’t, and a class set is a bad place to make them one script at a time. Three lines in the mark scheme settle most cases:

Mark each follow-through line as A1ft, and write “their” where a student’s own value is acceptable.
Say what happens to an impossible answer. The IB’s rule (method marks yes, answer mark no) is easy to apply.
Say whether a follow-through answer needs working. Edexcel’s rule (one step no, more than one step yes) is a fair default.

When you mark, the same slip usually turns up on several scripts. Work out the follow-through answer once for each common wrong value and write it in the margin of the mark scheme: for Question 1, gradient 3 gives $y = 3x$ or $y = 3x - 4$ . The tenth script then takes seconds.

Graidable is the tool we make. This part is about it.

How Graidable marks a carried error

Graidable drafts the marks on handwritten scripts from your mark scheme, and it follows the carried-error rule you write there. Where your scheme says nothing about carried errors, an error costs marks once and later work is judged against the student’s own value, as in Question 1. Each drafted mark shows where on the student’s page it comes from, and doubtful items carry a “Check closely” flag with the reason.

You approve every mark before anything reaches a student. Try the live demo and a sample class without a card, or see the subject pages for maths and chemistry.

Sources

AQA, GCSE Mathematics 8300 marking guidance, version 1.0, October 2023: mark types, “their”, follow-through marks (A1ft), misreads. Checked 8 October 2026. ↑
AQA, A-level Physics 7408/1 mark scheme, June 2018, section 3.4: errors carried forward, consequential marking and arithmetic errors. Checked 8 October 2026. ↑
Pearson, Edexcel GCSE Mathematics (1MA1) Paper 2F mark scheme, Summer 2025, general marking guidance, points 6 and 12 to 14, and the list of abbreviations. Checked 8 October 2026. ↑
OCR, A Level Physics A H556/01 Modelling physics mark scheme, June 2018, annotations AE, ECF, POT, TE and XP, and the guidance on question 16. Checked 8 October 2026. ↑
International Baccalaureate, Mathematics: analysis and approaches, Higher level, Paper 1 markscheme, November 2022 (N22/5/MATHX/HP1/ENG/TZ0/XX/M), “Instructions to Examiners”, sections 2, 4 and 5. The IB sells its markschemes, so there is no public link. Checked 8 October 2026. ↑

Sharing it with your department?

Guide · England and IB DP

How AQA, Edexcel, OCR and the IB mark an error carried forward

Every board lets a student keep the marks for correct work on a wrong earlier answer. They differ on the working they need to see and on where the allowance stops.

The Graidable team

Published 8 October 2026

Question 1: a straight line passes through $A(1, 3)$ and $B(5, 11)$ . Find (a) its gradient [2] and (b) its equation in the form $y = mx + c$ [2].

Student’s answer, Question 1

Marks, with ft

\text{(a)}\;\; m = \dfrac{11 - 3}{5 - 1}

m = \dfrac{8}{4} = \htmlClass{lab-err}{3}

A08 ÷ 4 = 2

(a) 1 [2]

\text{(b)}\;\; 11 = 3 \times 5 + c

M1uses own m

c = -4

y = 3x - 4

A1ft

(b) 2 [2]

Marking rule	(a) [2]	(b) [2]	Total [4]
With follow-through	1	2	3
Without	1	1	2

Figure 1. Constructed example. A slip in part (a), marked with follow-through: the error costs one mark, not two.

In this article

One error costs marks once
The four boards use different names for the same rule
Follow-through needs working on the page
Where follow-through stops
Write the rule into your own mark scheme

One error costs marks once

Here is a two-part question and its mark scheme.

Question 1 A straight line passes through $A(1, 3)$ and $B(5, 11)$ .

(a) Find the gradient of the line. [2]

(b) Find the equation of the line in the form $y = mx + c$ . [2]

Table 1. Mark scheme for Question 1
Part	Mark	What earns it
(a)	M1	$(11 - 3) / (5 - 1)$ , or an equivalent calculation
(a)	A1	2
(b)	M1	A point on the line and their gradient substituted into $y = mx + c$
(b)	A1ft	$y = 2x + 1$ , or the correct equation for their gradient and the point they used
Total	4

AQA’s marking guidance for GCSE Mathematics gives the reason in one line: the ft mark is there so that a student isn’t penalised twice, once for the earlier step and again for the final answer¹.

The four boards use different names for the same rule

Table 2. How each board writes and limits follow-through
Board and paper	Written as	What the board’s guidance says
AQA GCSE Mathematics	ft, A1ft	Method marks always follow a wrong value that came from a correct calculation. The final accuracy mark follows only where the scheme says A1ft¹.
AQA A-level Physics	ECF, conseq	Likely to be limited to calculation questions, and shown in the scheme. An arithmetic error costs one mark only².
Edexcel GCSE Mathematics	ft	Applies where the question’s own scheme allows it. A one-step follow-through can be credited without working; with more steps the working must be seen³.
OCR A Level Physics A	ECF	For numerical answers unless the scheme says otherwise. Given after arithmetic, transcription and power-of-ten errors, not after a wrong equation⁴.
IB Mathematics: analysis and approaches	FT	Applies between parts of a question, and working must normally be present. Inside a part, an error ends the accuracy marks but not the method marks⁵.

Follow-through needs working on the page

A wrong answer on its own is hard to follow through, because you can’t see that the student used their earlier value correctly. The boards handle this in slightly different ways.

Where follow-through stops

Follow-through rewards correct work on a wrong value. It was never meant to rescue wrong work, and each board says where the line is.

Question 2 A bag holds 5 red counters and 3 blue counters. Two counters are taken at random without replacement.

(a) Find the probability that both are red. [2]

(b) Find the probability that at least one is blue. [2]

Question 2: two counters are taken without replacement from 5 red and 3 blue. Find the probability that (a) both are red [2] and (b) at least one is blue [2].

Student’s answer, Question 2

Marks

\text{(a)}\;\; \dfrac{5}{8} \times \dfrac{4}{7}

= \htmlClass{lab-err}{\dfrac{20}{15}}

A08 × 7 = 56

(a) 1 [2]

\text{(b)}\;\; 1 - \dfrac{20}{15}

M1uses own value

= -\dfrac{5}{15}

A0not a probability

(b) 1 [2]

Figure 2. Constructed example. The method in part (b) follows through, but the answer cannot be a probability, so the accuracy mark is withheld.

Put together, the limits come down to four questions you can ask about any carried error (Figure 3).

Is the method in the later part itself wrong?
Yes: No follow-through. There is no correct work to credit.
No: go on to the next question
Was the correct value printed in the question, as in a “show that”?
Yes: No follow-through. The student should have used the given value.
No: go on to the next question
Is the answer impossible, such as a probability above 1?
Yes: Give the method marks and withhold the final answer mark.
No: go on to the next question
Did the error make the later part much easier?
Yes: Give fewer marks: only for the steps the student still had to do.
No: go to the last step
Otherwise: Follow through. Mark the later work as correct for the student’s own value.

Figure 3. Four checks before a carried error earns follow-through marks, drawn from the boards’ published guidance.

Write the rule into your own mark scheme

A board’s mark scheme arrives with these decisions made. Your own test doesn’t, and a class set is a bad place to make them one script at a time. Three lines in the mark scheme settle most cases:

Mark each follow-through line as A1ft, and write “their” where a student’s own value is acceptable.
Say what happens to an impossible answer. The IB’s rule (method marks yes, answer mark no) is easy to apply.
Say whether a follow-through answer needs working. Edexcel’s rule (one step no, more than one step yes) is a fair default.

Graidable is the tool we make. This part is about it.

How Graidable marks a carried error

You approve every mark before anything reaches a student. Try the live demo and a sample class without a card, or see the subject pages for maths and chemistry.

Sources

AQA, GCSE Mathematics 8300 marking guidance, version 1.0, October 2023: mark types, “their”, follow-through marks (A1ft), misreads. Checked 8 October 2026. ↑
AQA, A-level Physics 7408/1 mark scheme, June 2018, section 3.4: errors carried forward, consequential marking and arithmetic errors. Checked 8 October 2026. ↑
Pearson, Edexcel GCSE Mathematics (1MA1) Paper 2F mark scheme, Summer 2025, general marking guidance, points 6 and 12 to 14, and the list of abbreviations. Checked 8 October 2026. ↑
OCR, A Level Physics A H556/01 Modelling physics mark scheme, June 2018, annotations AE, ECF, POT, TE and XP, and the guidance on question 16. Checked 8 October 2026. ↑
International Baccalaureate, Mathematics: analysis and approaches, Higher level, Paper 1 markscheme, November 2022 (N22/5/MATHX/HP1/ENG/TZ0/XX/M), “Instructions to Examiners”, sections 2, 4 and 5. The IB sells its markschemes, so there is no public link. Checked 8 October 2026. ↑

Sharing it with your department?