The single habit that separates students who get unstuck on proofs from students who sit staring at the page is almost embarrassingly simple: keep two pieces of paper. One is scratch, where you run the logic in whichever direction makes progress — usually backwards from the conclusion. The other is the fair copy, where the final proof has to run forwards from the hypothesis. The two columns do different jobs, and confusing them is what makes proofs feel impossible.

This article is not about any particular proof. It is about the workflow that turns a blank page into a finished argument — a routine you can apply to every direct proof in your homework, every week, without thinking about which technique to use.

The two columns

Imagine your desk split down the middle. On the left, scratch paper. On the right, the fair copy.

The scratch paper is where you think. The fair copy is where you write. These are different activities, and collapsing them into one leads to two symmetric failures: you either try to write a clean proof before you understand it (and get stuck on line three), or you turn in a scratch sheet (and the grader cannot follow the argument).

The recipe

Here is the routine, for any direct proof "if P then Q."

On scratch paper:

  1. Write P at the top. Write Q at the bottom. Leave space between them.
  2. Unpack P using its definitions — write whatever that unlocks just below P.
  3. Unpack Q using its definitions — write what the conclusion will need to look like, just above Q.
  4. Now work the gap. Try forwards from P's unpacking and backwards from Q's unpacking. Write arrows pointing both ways. Mark any step that is not reversible with a warning sign (see Working Backwards).
  5. When the two halves meet, you have a route. Box the sequence of intermediate statements.

On the fair copy:

  1. Rewrite the boxed sequence, running from P at the top to Q at the bottom. Every step gets a justification (by definition, by algebra, by a previously proved theorem). Use because for assertions, if only for genuine hypothetical scopes (see Because vs If).
  2. End with a \blacksquare or QED so the reader knows the argument is done.

The handoff from step 5 to step 6 is the whole trick. On the scratch sheet the argument bends and circles; on the fair copy it runs in one direction.

A picture of the handoff

Scratch paper and fair copy, side by sideTwo panels side by side. The left panel, labelled scratch, shows arrows going from the hypothesis P downwards and also from the conclusion Q upwards, meeting in the middle with a boxed route. The right panel, labelled fair copy, shows the same route written as a clean top-to-bottom chain from P to Q with no return arrows. A draggable dot highlights the path moving from the scratch layout to the fair-copy layout. scratch P unpack P unpack Q Q fair copy P R₁ (because P) R₂ (because R₁) Q ∎ drag to watch the route move from scratch to fair copy
Scratch (left) has the route running inwards from both ends, with an arrow from "unpack $Q$" back upwards toward "unpack $P$" marked in red to show it was found by working backwards. Fair copy (right) has the same route, rewritten to run forwards, every step justified by the line above it.

A tiny worked example

Claim. If n is an integer, then n(n+1) is even.

Scratch. I want n(n+1) even, i.e. n(n+1) = 2k for some integer k. Case split: n even or n odd. If n even, n = 2m, then n(n+1) = 2m(n+1) — done. If n odd, then n+1 is even, n+1 = 2m, then n(n+1) = n \cdot 2m = 2nm — done. Route found.

Fair copy. If n is even, write n = 2m for some integer m. Then n(n+1) = 2m(n+1), which is 2 times an integer, so n(n+1) is even. If n is odd, then n+1 is even (consecutive integers alternate parity), so n+1 = 2m for some integer m. Then n(n+1) = n \cdot 2m = 2(nm), which is 2 times an integer, so n(n+1) is even. In either case, n(n+1) is even. \blacksquare

The scratch sheet was one line. The fair copy is three times longer — but every sentence in the fair copy does a job the scratch line does not: it names the case, cites the definition, and makes the conclusion explicit. The scratch sheet found the route; the fair copy communicates it.

What this habit buys you

Three things.

1. You stop getting stuck. A blank page is intimidating when your only allowed move is "write the first line of the final proof." A scratch sheet lets you write anything — guesses, rearrangements, backward chains. Once the route is visible, writing the final version is mechanical.

2. You catch your own errors. The two directions give you two chances to check each step. If the scratch chain claims A \Leftrightarrow B but the forward version only shows A \Rightarrow B, you catch it during the handoff. Mistakes that would slip past a single-pass proof get caught by the flip.

3. You stop believing textbook proofs came out of nowhere. Every textbook proof was a scratch sheet once. Seeing your own scratch-sheet-plus-fair-copy workflow demystifies the published version. You are not missing some genius instinct; the textbook author just did not show you the left column.

A diagnostic checklist

Before handing in a proof, run these three checks:

Three yeses means the two-column handoff worked. You have a proof that was discovered by the most effective method available, and presented in the form the reader needs.

The short summary

If you build this habit, the question "where do I start?" disappears. You start on the scratch paper, always, and the fair copy writes itself once the route is found.

Related: Mathematical Proof — Direct Proof · Scratch Work vs Final Proof · Working Backwards: Cheating or Not? · Because vs If Inside a Proof