Coding–Decoding

In short: The alphabet-position table + every coding model: number, letter, symbol, substitution, conditional.

Reasoning — SI › Verbal & Analytical Reasoning · 42 min read · Free to read · English & Telugu

Bilingual study notes for the Andhra Pradesh Police SI & Constable exams (SLPRB AP) by AP Police Hub, a product of Anvil Academy.

Number, letter, symbol, substitution, sentence and conditional codes — the complete SI ladder, built on the alphabet-position table, with every model worked step by step.

1. Introduction & why it matters

Coding–Decoding is the art of hiding a word or message behind a secret rule and then recovering it. The sender codes the message; the receiver decodes it. Your only job in the exam is to discover the hidden rule from the example given, and then apply that same rule to a new word. In the APSLPRB SI Prelims this is one of the most reliable scoring areas in the Reasoning section — worth a steady 3–6 marks — because every question is rule-based and, once the models are drilled, solvable in under a minute.

2. The alphabet-position table — your master tool

The single most powerful aid in coding is the position of each letter in the English alphabet. Learn it both ways — forward (A = 1 … Z = 26) and backward / reverse (Z = 1 … A = 26). If you can place any letter instantly, three-quarters of the models below collapse into simple arithmetic.

Forward position (A = 1 … Z = 26)
A	B	C	D	E	F	G	H	I	J	K	L	M
1	2	3	4	5	6	7	8	9	10	11	12	13

N	O	P	Q	R	S	T	U	V	W	X	Y	Z
14	15	16	17	18	19	20	21	22	23	24	25	26

Reverse position (Z = 1 … A = 26)

The reverse rank — position counted from the end (Z-side)
A	B	C	D	E	F	G	H	I	J	K	L	M
26	25	24	23	22	21	20	19	18	17	16	15	14

N	O	P	Q	R	S	T	U	V	W	X	Y	Z
13	12	11	10	9	8	7	6	5	4	3	2	1

3. Number coding

In number coding each letter is replaced by a number. The number can be arbitrary (a direct code you must deduce), the letter's alphabet position, a sum of positions, or a value read off a supplied table. Four models cover them all.

Model 1 — Direct number coding (deduce the code)

The numbers follow no positional logic — each letter simply owns a fixed digit. Line the letters of a coded word beneath the word, read off the value of each letter, then reuse those values on the target word.

Model 1

Question: In a code language, 'FLAME' is written as '3-8-2-5-7' and 'LEAF' is written as '8-7-2-3'. How is 'MEAL' written in that code?

Approach: The digits carry no positional logic here, so each letter simply owns a fixed number that we must read off. Line each word up against its code letter by letter to build the map, and use the letters the two words have in common as a check that the map is right.

F\to3,\ L\to8,\ A\to2,\ M\to5,\ E\to7 — Place the five letters of FLAME beneath its code 3\text{-}8\text{-}2\text{-}5\text{-}7 to read the value of each:
L\to8,\ E\to7,\ A\to2,\ F\to3\ \checkmark — LEAF shares the letters L, E, A, F with FLAME, so its code must reuse the same values — and 8, 7, 2, 3 matches exactly, confirming the map is trustworthy:
M\to5,\ E\to7,\ A\to2,\ L\to8 — Now spell out MEAL using those same fixed values, letter by letter:

Answer: Reading the fixed codes gives MEAL = 5-7-2-8.

Shortcut: Always confirm the code on a SECOND given word before trusting it — shared letters are your proof.

Model 2 — Position-in-alphabet coding

Each letter is replaced by its alphabet position. Two flavours appear: the forward position (A = 1) and the reverse position (Z = 1). Spot which one from the example, then apply it.

Model 2a — forward

Question: If 'SIGN' is coded as '19-9-7-14', how is 'RING' coded in the same language?

Approach: Before assuming the code is arbitrary, always test whether the numbers are simply the letters' alphabet positions. The very first letter settles it: S codes to 19, which is exactly where S sits in the alphabet, so this is plain forward-position coding.

Verify on the rest of SIGN: I=9, G=7, N=14 all agree with their forward positions, so the rule is confirmed for every letter:
R=18,\ I=9,\ N=14,\ G=7 — Now replace each letter of RING by its own forward position:

Answer: Using forward positions, RING = 18-9-14-7.

Model 2b — reverse

Question: Using reverse positions (Z = 1, A = 26), how is 'CAT' coded?

Approach: Reverse position counts from the far end, Z=1 up to A=26. Rather than counting backwards for each letter, use the neat fact that a letter's forward and reverse positions always add to 27, so the reverse position is just 27 minus the forward one.

C:\ 27-3=24 — C has forward position 3, so its reverse position is 27-3:
A:\ 27-1=26 — A has forward position 1, so its reverse position is 27-1:
T:\ 27-20=7 — T has forward position 20, so its reverse position is 27-20:

Answer: Reading reverse positions, CAT = 24-26-7.

Shortcut: Reverse position is just the opposite letter's forward position: C↔X=24, A↔Z=26, T↔G=7.

Model 3 — Adding up positional values (word → number)

Here the whole word becomes one number. The commonest rule is the plain sum of positions, but SI papers love a twist: multiply the sum by the number of letters, or add/subtract the letter-count, or (surprisingly) code by the count of letters alone. Always test the plain sum first, then look for the twist.

Model 3a — plain sum

Question: If 'SUN' is coded as 54, how is 'MOON' coded by the same rule?

Approach: When a whole word is coded as a single number, the first rule to try is always the plainest one: add up the alphabet positions of all its letters. Test it on the given word, and if the total lands on the given code, apply the same sum to the target word.

S+U+N = 19+21+14 = 54 — Add the positions of S, U, N. The total is 54, exactly the code given for SUN, so the rule really is the plain sum of positions:
M+O+O+N = 13+15+15+14 = 57 — Now add the positions of the four letters of MOON the same way:

Answer: The sum of MOON's positions is 57, so MOON = 57.

Model 3b — sum × letter-count

Question: If 'SHE' is coded as 96, how is 'ME' coded in the same system?

Approach: Start as always with the plain sum, but here it comes to 32 while the code is 96 — three times as big. When the code is a clean multiple of the sum, the missing factor is usually something about the word itself, so compare that factor with the number of letters.

S+H+E = 32 — First add the positions of SHE, which come to 32 — well short of the code 96:
32 \times 3 = 96\ \checkmark — Divide the code by the sum: 96 \div 32 = 3, and SHE happens to have exactly 3 letters — so the rule is 'sum \times number of letters':
18 \times 2 = 36 — For ME the sum is 13 + 5 = 18 and it has 2 letters, so multiply the two:

Answer: Sum 18 times its 2 letters gives ME = 36.

Model 3c — the letter-count trap

Question: If 'EAGLE' is coded as 125, how is 'SUN' coded in the same code?

Approach: The sum of EAGLE's positions is far below 125, so a positional rule cannot be behind this code. When a short word produces a surprisingly large number, suspect that the rule ignores the letters entirely and works on the letter-count instead.

5^3 = 125\ \checkmark — EAGLE has 5 letters, and 5^3 = 125 — the code is simply the cube of the number of letters:
3^3 = 27 — SUN has 3 letters, so cube that count:

Answer: The code is the cube of the letter-count, so SUN = 3^3 = 27.

Shortcut: When a small word gives a big code, suspect a rule based on the LETTER-COUNT, not the letters.

Model 4 — Coding from a supplied letter ↔ number table

Sometimes the question simply hands you a table of codes. There is no rule to crack — just read the value of each letter and write the word out.

Given code table
Letter	P	L	A	N	E	T
Code	5	2	8	6	3	9

Model 4

Question: Using the table above, what is the code for the word 'PLANET'?

Approach: There is no hidden rule to crack here — the table already hands you the code for every letter. The only work is careful reading: take the letters of the word in order and copy each one's value straight from the table.

P\to5,\ L\to2,\ A\to8,\ N\to6,\ E\to3,\ T\to9 — Take PLANET one letter at a time and read each code from the table in order:

Answer: Reading the table in order, PLANET = 5-2-8-6-3-9.

4. Letter coding

In letter coding a word is coded by other letters. The replacement may be a fixed substitution, a uniform jump forward or back along the alphabet, the opposite letter, or a rearrangement of the same letters. These are the highest-frequency SI models — master all four.

Model 5 — Direct letter substitution (deduce the rule from two words)

Each letter has a fixed but arbitrary substitute (no shift pattern). Two coded words are given; wherever they share a letter, its code must match — that agreement lets you pin down the substitution and code a third word.

Model 5

Question: If 'TRAIN' is coded as 'MPESQ' and 'PAINT' is coded as 'HESQM', how is 'PRINT' coded in the same language?

Approach: Each letter has its own fixed substitute with no shift pattern, so the only way in is the letters the two given words share. Line each word up with its code, note the substitute for every letter, and let the overlap between the words confirm the map before you build the third word.

T\to M,\ R\to P,\ A\to E,\ I\to S,\ N\to Q — Align TRAIN with its code MPESQ, letter under letter, to read five substitutions:
P\to H\ \checkmark — PAINT and TRAIN share A, I, N, T; aligning PAINT with HESQM shows those four match perfectly and adds the new pairing P\to H, so the map is trustworthy:
P\to H,\ R\to P,\ I\to S,\ N\to Q,\ T\to M — Every letter of PRINT is now known, so substitute them one by one:

Answer: Substituting each known letter, PRINT = HPSQM.

Model 6 — Alphabet jump (+n / −n shift)

Every letter moves the same number of steps forward (+n) or backward (−n). Find the gap on one letter, confirm on a second, then apply. The one trap: when a step runs past Z, it wraps around to A (and past A back to Z).

Model 6a — forward shift

Question: If 'MOBILE' is coded as 'PRELOH', how is 'PHONE' coded in the same code?

Approach: In a shift code every letter moves the same number of steps along the alphabet. Measure that step on the first letter — M becomes P, a jump of +3 — then confirm the very same jump holds for the rest of the word before you trust it.

\text{shift} = +3 — Check the jump on every letter of MOBILE: M{+}3{=}P, O{+}3{=}R, B{+}3{=}E, I{+}3{=}L, L{+}3{=}O, E{+}3{=}H — a clean, uniform +3:
P{+}3{=}S,\ H{+}3{=}K,\ O{+}3{=}R,\ N{+}3{=}Q,\ E{+}3{=}H — Move each letter of PHONE forward by the same 3 places:

Answer: Shifting every letter +3 gives PHONE = SKRQH.

Model 6b — backward shift with wrap

Question: If 'FROG' is coded as 'COLD', how is 'BAND' coded in the same code?

Approach: This is a backward shift — F becomes C, a jump of -3. Backward codes carry one hazard: for letters near the start of the alphabet the count runs off the front, so it wraps around from A back to Z. Handle those letters by adding 26 to keep the position in range.

\text{shift} = -3 — Confirm the shift on FROG: F{-}3{=}C, R{-}3{=}O, O{-}3{=}L, G{-}3{=}D — a steady -3:
B\ (2)\ \to\ 2-3=-1\ \to\ +26 = 25 = Y — B is at position 2; subtracting 3 gives -1, which is off the front, so add 26 to wrap it round to position 25 = Y:
A\to X,\ N\to K,\ D\to A — A also wraps (to X), while N and D sit far enough along to shift back normally:

Answer: Shifting -3 with wrap-around, BAND = YXKA.

Shortcut: To go back past A, add 26: (position − n + 26). Going forward past Z, subtract 26.

Model 7 — Opposite-letter coding (the 27 − n rule)

Each letter is replaced by its opposite — the letter equidistant from the other end (A↔Z, B↔Y, …). The code letter's position is 27 − (original position). This is the same as reading the reverse-position table.

Model 7

Question: If 'COPPER' is coded as 'XLKKVI', how is 'SILVER' coded in the same code?

Approach: In opposite-letter coding each letter is swapped for the one equidistant from the other end of the alphabet — A with Z, B with Y, and so on. You can read these straight off the opposite pairs, or compute each one as 27 minus its position.

First confirm the rule on COPPER: its opposites C{\leftrightarrow}X, O{\leftrightarrow}L, P{\leftrightarrow}K, P{\leftrightarrow}K, E{\leftrightarrow}V, R{\leftrightarrow}I spell exactly XLKKVI:
S\to H,\ I\to R,\ L\to O,\ V\to E,\ E\to V,\ R\to I — Now replace each letter of SILVER by its opposite pair:
27-19 = 8 = H — As a safety check, verify one letter by the formula: S is at position 19, and 27-19 = 8 = H, matching the pair above:

Answer: Replacing each letter by its opposite, SILVER = HROEVI.

Model 8 — Jumbled-letter coding (rearranging the same letters)

No letter changes — only their order does. The three exam patterns are: full reversal, swapping adjacent pairs (1↔2, 3↔4, …), and reversing each half of the word.

Model 8a — full reversal

Question: If 'RESPONSIBLE' is coded as 'ELBISNOPSER', how is 'ENJOYMENT' coded in the same code?

Approach: None of the letters have changed — only their order has — which points to a jumbling code rather than a substitution. Compare the first letters: the code begins with the word's last letter, the plain sign that the whole word has simply been written backwards.

Read ELBISNOPSER backwards and it is exactly RESPONSIBLE, so the rule is simply 'write the word in reverse':
E\,N\,J\,O\,Y\,M\,E\,N\,T \to T\,N\,E\,M\,Y\,O\,J\,N\,E — Apply the same rule to ENJOYMENT, writing its letters from last to first:

Answer: Reversing the word gives ENJOYMENT = TNEMYOJNE.

Model 8b — swap adjacent pairs

Question: If 'MATTER' is coded as 'AMTTRE', how is 'LENGTH' coded in the same code?

Approach: The letters are the same but not fully reversed, so look more closely at how they have moved. Here they trade places two at a time — the 1st with the 2nd, the 3rd with the 4th, and so on — a swapping-of-adjacent-pairs code.

Break MATTER into pairs and swap within each: MA\to AM, TT\to TT, ER\to RE, which spells AMTTRE — confirming the pair-swap rule:
(L\,E)(N\,G)(T\,H) \to (E\,L)(G\,N)(H\,T) — Do the same to LENGTH — split it into three pairs and swap the two letters in each:

Answer: Swapping each adjacent pair, LENGTH = ELGNHT.

Model 8c — reverse each half

Question: If 'GARDEN' is coded as 'RAGNED', how is 'PENCIL' coded in the same code?

Approach: Again the letters only change order, but this time neither a full reversal nor a pair-swap fits. The trick is to cut the word down the middle into two equal halves and reverse the letters inside each half on its own.

Split GARDEN into GAR | DEN and reverse each half to get RAG | NED = RAGNED, which is the given code — so the rule is 'reverse each half':
PEN \to NEP,\quad CIL \to LIC — Cut PENCIL the same way into PEN | CIL and reverse the letters in each half:

Answer: Reversing each half, PENCIL = NEPLIC.

5. Mixed & mathematical coding

The harder SI codes combine two operations: rearrange the letters and then shift them, or treat vowels and consonants by separate rules. Always undo or trace the operations in the right order.

Model 9 — Reverse, then shift

First the word is written backwards, then a uniform shift is applied to every letter. Discover the shift by lining up the reversed word against the code.

Model 9

Question: If 'STRONG' is coded as 'HOPSUT', how is 'PLANET' coded in the same code?

Approach: This code hides two steps stacked on top of each other. When neither a plain shift nor a plain reversal matches, try reversing the word first and only then lining it up against the code; the leftover difference reveals the uniform shift applied afterwards.

STRONG \to GNORTS — First reverse STRONG to get GNORTS, undoing the jumbling part of the code:
\text{reverse, then }+1 — Set GNORTS against the code HOPSUT: every letter is one step ahead (G{+}1{=}H, N{+}1{=}O, and so on), so the second operation is a uniform +1:
T{+}1{=}U,\ E{+}1{=}F,\ N{+}1{=}O,\ A{+}1{=}B,\ L{+}1{=}M,\ P{+}1{=}Q — Apply the same two steps to PLANET: reverse it to TENALP, then push each letter forward by 1:

Answer: Reversing then shifting +1, PLANET = UFOBMQ.

Model 10 — Operation-then-code: vowels & consonants split

Vowels obey one rule and consonants another. Mark the vowels first, then apply each rule separately.

Model 10

Question: If 'GARDEN' is coded as 'FBQCFM', how is 'PLANET' coded in the same code?

Approach: If a single shift does not fit the whole word, the letters may be moving by two different rules. The usual split is vowels against consonants, so mark the vowels of the example first and watch whether the two groups travel in opposite directions.

\text{consonant} \to -1 — Look at the consonants of GARDEN — G, R, D, N. Each has moved back one place (G{\to}F, R{\to}Q, D{\to}C, N{\to}M), so consonants follow -1:
\text{vowel} \to +1 — The vowels A and E have instead moved forward one place (A{\to}B, E{\to}F), so vowels follow +1 — the two groups go opposite ways:
P{-}1{=}O,\ L{-}1{=}K,\ A{+}1{=}B,\ N{-}1{=}M,\ E{+}1{=}F,\ T{-}1{=}S — Apply both rules to PLANET at once — vowels A, E forward, consonants P, L, N, T back:

Answer: Vowels +1 and consonants -1 give PLANET = OKBMFS.

6. Symbol & substitution coding

Model 11 — Symbol coding

Letters are replaced by symbols (@, #, %, & …). It works exactly like direct letter coding — build the letter-to-symbol map from the given words, then write the target word in symbols.

Model 11

Question: If 'TRUCK' is written as '@#%&*' and 'CUT' is written as '&%@', how is 'CURT' written in that code?

Approach: Symbols behave exactly like the letters of a direct code — each letter simply owns one fixed symbol. Build the letter-to-symbol map by lining each word up with its code, and use the letters the two words have in common to check the map is right.

T\to\text{@},\ R\to\#,\ U\to\%,\ C\to\&,\ K\to* — Place the five letters of TRUCK beneath its symbols to read one symbol per letter:
CUT shares C, U, T with TRUCK; its code \&\%\text{@} uses exactly those symbols, so the map holds:
C\to\&,\ U\to\%,\ R\to\#,\ T\to\text{@} — Now write CURT by reading each of its letters' symbols in order:

Answer: Reading each symbol, CURT = &%#@.

Model 12 — Substitution coding (things called other names)

Everyday words are renamed. To answer, find the real object the question asks for, then report what that object is called in the code.

Model 12

Question: If 'water' is called 'milk', 'milk' is called 'oil', and 'oil' is called 'honey', then what do we drink to quench thirst?

Approach: Substitution puzzles work in two moves, and the mistake to avoid is answering with the real object. First settle what the real answer is in ordinary life, then translate that object into whatever the code calls it.

In ordinary life thirst is quenched by drinking water, so the real answer is 'water':
\text{water} \to \text{milk} — But the code renames things, and here 'water' is called 'milk', so that is the name to give:

Answer: We really drink water, but in this code it is called 'milk'.

Shortcut: Answer in the code's language, never the real word — the code name is what they want.

7. Sentence / message coding

Whole sentences are coded, word for word, but the codes are scrambled (not in order). To find one word's code, pick sentences that share exactly one word — the code they share must be that word.

Model 13

Question: In a code language: 'roses are red flowers' = 'la ka de ho'; 'red pens are costly' = 'de mi ka zo'; 'beautiful red roses' = 'la de fu'. Find the code for 'roses'.

Approach: The codes in each sentence are scrambled, so you cannot match word to code by position. The way in is overlap: a word that appears in several sentences must be coded by the one symbol those sentences all share. Pin down such a word first, then use it to narrow the rest.

\text{red} \to \text{de} — The word 'red' is the only word present in all three sentences, and 'de' is the only code common to all three — so 'red' must be 'de':
\{la, ka, de, ho\} \cap \{la, de, fu\} = \{la, de\} — 'roses' appears in sentences 1 and 3, so its code lies in what those two share — the common codes \{la, de\}:
Of those two, 'de' is already taken by 'red', so the only choice left for 'roses' is 'la':

Answer: After removing the code for 'red', the word 'roses' must be 'la'.

Shortcut: Two sentences sharing one word share one code. Cross out codes you have already identified.

8. Conditional coding

The hardest SI model. A fixed code table (letters → numbers/symbols) is given, plus conditions that override the table when a word starts or ends with a vowel or consonant. Read the conditions in order, apply the first one that fits, and code the middle letters normally.

Code table (letter → code)
Letter	A	D	E	G	H	I	N	R	T
Code	1	@	2	#	3	4	5	%	6

Model 14a — condition (I)

Question: Using the table and conditions, code the word 'HATE'.

Approach: The conditions only ever depend on the first and last letters, so always classify those two before anything else. HATE begins with the consonant H and ends with the vowel E, which is exactly the pattern condition (I) describes.

H \to *,\quad E \to * — Since the first is a consonant and the last a vowel, condition (I) fires and both end letters are coded '*', overriding the table:
A \to 1,\quad T \to 6 — The middle letters are untouched by any condition, so read them straight from the table — A=1, T=6:

Answer: With condition (I) on the ends, HATE = *16*.

Model 14b — condition (II)

Question: Using the same table and conditions, code the word 'IDEA'.

Approach: Check the two end letters first, as always. IDEA opens with the vowel I and closes with the vowel A, so both ends are vowels — the situation condition (II) is written for, which tells us to interchange their table codes.

I \to 1,\quad A \to 4 — From the table I = 4 and A = 1; condition (II) swaps these two values, so I takes 1 and A takes 4:
D \to \text{@},\quad E \to 2 — The inner letters are coded normally from the table — D=@, E=2:

Answer: Swapping the vowel end-codes, IDEA = 1@24.

Model 14c — condition (III)

Question: Using the same table and conditions, code the word 'IRAN'.

Approach: Classify the ends of IRAN: it starts with the vowel I and ends with the consonant N. That is the reverse of condition (I), and it is condition (III) that covers it — both end letters take the code of the first letter.

I \to 4,\quad N \to 4 — Condition (III) fires: both the first and last letters are coded with the first letter's value, and I = 4, so both become 4:
R \to \%,\quad A \to 1 — The middle letters follow the plain table — R=\%, A=1:

Answer: Both ends take I's code under condition (III), so IRAN = 4%14.

Shortcut: Always test the first and last letters against the conditions BEFORE touching the middle.

9. Decoding — working the rule backwards

So far we coded a word. The reverse task — decoding — gives you a code and asks for the original word. The method is identical: find the rule, then undo it. A +n shift is undone by −n; a table look-up is read the other way.

Model 15a — decode a shift code

Model 15a

Question: In a code, each letter is replaced by the next letter (+1). The code 'CBU' stands for which word?

Approach: Decoding is coding run in reverse. The word was coded by moving each letter forward one place (+1), so to recover the original you must undo that step and move each code letter back one place (-1).

C{-}1{=}B,\ B{-}1{=}A,\ U{-}1{=}T — Step each letter of the code CBU back by one place to reverse the +1 rule:

Answer: Undoing the +1 shift, 'CBU' decodes to BAT.

Shortcut: Verify by re-coding: BAT +1 = CBU. ✓

Model 15b — decode from a number ↔ letter table

A table pairs numbers with letters; a group of numbers must be turned back into a meaningful word. Read each number's letter and reassemble.

Given number ↔ letter table
Number	6	8	5	2	0	1	4
Letter	B	R	A	I	N	D	T

Model 15b

Question: Using the table above, what meaningful word does the code '48520' represent?

Approach: Here the table pairs numbers with letters, and decoding just means reading that table the other way. Take the code digit by digit, look up the letter for each, and the letters should fall into a meaningful word.

4\to T,\ 8\to R,\ 5\to A,\ 2\to I,\ 0\to N — Look up each digit of 48520 in the table and write down its letter:
T\,R\,A\,I\,N — Read the letters in order and they spell a real word:

Answer: Reading the table backwards, '48520' decodes to TRAIN.

10. Shortcuts, traps & tips

11. At a glance

Every model, its clue and its method
Model	Tell-tale clue	Method
Direct number/letter/symbol	No pattern in the codes	Map each letter from two words; reuse
Position coding	A=1 (or Z=1) shows up	Replace by forward or reverse position
Positional sum	A word becomes ONE number	Sum positions; test ×, ÷, ± letter-count
Alphabet jump	Letters shift by a fixed n	Find +n / −n; mind the wrap
Opposite letter	A↔Z, B↔Y look-alikes	Replace by 27 − position
Jumbled letters	Same letters, new order	Reverse / swap pairs / reverse halves
Reverse-then-shift	Two operations at once	Reverse the word, then apply the shift
Vowel/consonant split	Vowels move differently	Mark vowels; apply each rule apart
Substitution	'X is called Y'	Find real object, report its code name
Sentence coding	Word-for-word, scrambled	Share-one-word → share-one-code
Conditional	A code table + condition list	Check first/last letters; then table
Decoding	Code given, word wanted	Find the rule, then undo it