In a certain code language, "FIGURE" is written as "GKUGSG" and "TRIPOD" is…

In this letter-coding family, a six-letter word is split into three two-letter blocks — positions 1-2, 3-4 and 5-6. The two outer blocks are enciphered by shifting the block's first letter one place forward in the alphabet and its second letter two places forward; the middle block is left un-shifted but its two letters are exchanged with each other.

1. FIGURE splits into the blocks FI, GU and RE, and its code GKUGSG splits into GK, UG and SG.

2. Outer block FI: F shifts one place to G and I shifts two places to K, giving GK — matches. Outer block RE: R shifts one place to S and E shifts two places to G, giving SG — matches.

3. Middle block GU simply swaps to UG — matches, confirming the middle block is exchanged, not shifted.

4. The same rule holds for TRIPOD to UTPIPF: blocks TR, IP, OD code to UT (T+1, R+2), PI (IP swapped), and PF (O+1, D+2).

5. Apply the confirmed rule to LARGER, split into LA, RG and ER: the outer blocks shift (L+1=M, A+2=C) and (E+1=F, R+2=T), while the middle block RG simply swaps to GR.

6. Combining the three coded blocks MC, GR and FT gives the coded word MCGRFT.

Reversing the same rule on MCGRFT recovers the original word: undo the outer-block shifts (M-1=L, C-2=A) and (F-1=E, T-2=R), and undo the middle-block swap (GR back to RG), which reassembles LA-RG-ER = LARGER, confirming the coding is consistent both ways.