Recurring slips in GATE questions, by subject

Every solved question in the IAE study library ends with a note on the slip that question can lead to. In the 17 subjects this post covers, the 1,554 notes show the same slips coming back from question to question — a polarity read the wrong way round, a factor of 2π left out, a word chosen for its sound rather than its meaning. This post lists the recurring slips found in each subject — up to eight per subject, 120 in all — each recorded in at least three of that subject’s worked solutions and linked to up to three past questions that turn on it, for a candidate checking their own working against the slips recorded on past questions. The notes are our own view of where each question can go wrong. They are not a record of what candidates got wrong.

Subjects in this post

• General Aptitude (EC, EE and IN) — 8 slips
• Engineering Mathematics (EC, EE and IN) — 8 slips
• Networks (Electric Circuits) (EC, EE and IN) — 8 slips
• Signals and Systems (EC, EE and IN) — 8 slips
• Control Systems (EC, EE and IN) — 8 slips
• Analog Electronics (EC, EE and IN) — 8 slips
• Digital Electronics (EC, EE and IN) — 8 slips
• Electronic Devices (EC) — 4 slips
• Electromagnetics (EC, EE and IN) — 8 slips
• Electrical Machines (EE and IN) — 6 slips
• Electrical and Electronic Measurements (EE and IN) — 6 slips
• Programming in C (CS) — 4 slips
• Data Structures and Algorithms (CS) — 7 slips
• Computer Organization and Architecture (CS) — 7 slips
• Fundamentals of Biological Engineering (BT) — 7 slips
• Genetics, Cell and Molecular Biology (BT) — 7 slips
• Recombinant DNA Technology and Tools (BT) — 8 slips

What the counts mean

Each subject lists up to eight slips, each noted in at least three of its worked solutions; a slip beyond a subject’s first eight, or noted in fewer than three, is not listed. Each count is a minimum, which is why it says “at least”: a note describing the same slip in other words may not be counted. A count is out of all the subject’s worked solutions, many of them questions where that slip could not arise, so it is not a rate. A question can count towards more than one slip, and a slip that turns up in several subjects is counted in each. Within a subject, slips are ordered by count.

The count, subject by subject

General Aptitude and Engineering Mathematics come from the EC, EE and IN papers, like the other electrical and electronics subjects; the CS subjects come from ten CS papers and the BT subjects from seven BT papers. The subjects differ in size — from 34 worked solutions for Programming in C to 206 for General Aptitude — so their counts are not comparable with each other: a count says how many of that subject’s solutions record a slip, nothing more.

General Aptitude (EC, EE and IN)

• Choosing by sound, spelling or prefix instead of meaning — Picking a word because it resembles the right one in sound, spelling or prefix, rather than testing what it means or what relation the pair holds. Recorded in at least 15 of the 206 worked solutions; for example EE 2026 Q.6, IN 2026 Q.6 and EC 2025 Q.1.
• Testing only one case, value or condition before accepting — Checking a single column, value, example or condition and stopping once it works. A wrong option can pass that one test and fail the next. Recorded in at least 13 of the 206 worked solutions; for example EC 2026 Q.2, IN 2025 Q.3 and EE 2022 Q.6.
• Reporting a related quantity instead of the one asked — Stopping at an intermediate value, or answering with a nearby quantity such as a head start, a half-length or another person's figure, instead of the final quantity the question names. Recorded in at least 10 of the 206 worked solutions; for example EE 2025 Q.5, EC 2022 Q.3 and IN 2018 GA Q.4.
• Misreading what the question asks for — Answering a different question from the one printed: choosing the correct statement when the incorrect one is asked, the wrong order, or a related count or average in place of the one named. Recorded in at least 8 of the 206 worked solutions; for example EC 2025 Q.9, EE 2025 Q.8 and IN 2018 GA Q.5.
• Accepting a conclusion the passage does not state — Treating an overreach as supported: stretching a narrow claim into a broad one, importing outside knowledge or opinion, or hardening a worry into a certainty. The passage itself does not say it. Recorded in at least 7 of the 206 worked solutions; for example EE 2026 Q.4, IN 2025 Q.6 and EC 2022 Q.6.
• Measuring from a figure that is not drawn to scale — Reading lengths, positions or shapes straight off a drawing that is only representative or qualitative. The answer has to come from the stated data, and the picture can mislead. Recorded in at least 7 of the 206 worked solutions; for example EE 2026 Q.7, EC 2023 Q.7 and IN 2025 Q.9.
• Sign slips when rearranging an equation — A plus or minus sign is lost while moving terms, expanding a product or applying a formula, which gives wrong roots or a different form of the equation. Recorded in at least 6 of the 206 worked solutions; for example EE 2026 Q.3, IN 2026 Q.10 and EC 2020 GA Q.9.
• Off-by-one counting of steps, days or cycle positions — Counting items where the question needs the gaps between them, or the reverse, and so landing one too high or too low. A remainder of zero is also placed at the wrong point of a cycle. Recorded in at least 5 of the 206 worked solutions; for example IN 2026 Q.4, EC 2023 Q.10 and EE 2023 Q.9.

Engineering Mathematics (EC, EE and IN)

• Reporting a nearby quantity instead of the one asked — Computing something close to what the question wants, such as the minimiser rather than the minimum, the squared quantity, a slope rather than an area, or a joint rather than a conditional probability. Recorded in at least 9 of the 188 worked solutions; for example IN 2026 Q.31, EC 2025 Q.59 and EE 2023 Q.34.
• Dropping a constant multiplier or symmetry factor — Leaving out a numerical factor such as 2π, a symmetry multiple, a half or a factorial, so the result is scaled wrongly even though the method is sound. Recorded in at least 8 of the 188 worked solutions; for example EE 2026 Q.65, IN 2026 Q.13 and EC 2022 Q.53.
• Wrong sign on a residue or the contour direction — Losing a minus sign while evaluating a residue, or reading the direction of the contour the wrong way round, so the integral comes out with the wrong sign or cancels when it should not. Recorded in at least 7 of the 188 worked solutions; for example IN 2025 Q.33, EC 2022 Q.32 and EE 2020 Q.5.
• Treating a necessary condition as if it were sufficient — Taking a condition that must hold, such as terms tending to zero or no two rows being parallel, as proof of the conclusion, when further checks are still needed. Recorded in at least 6 of the 188 worked solutions; for example EE 2026 Q.49, EC 2025 Q.12 and IN 2025 Q.49.
• Counting a pole that lies outside the contour — The slip is including the residue of a pole that sits outside the contour (or adding both residues when only one pole is enclosed), which changes the value of the integral. Recorded in at least 5 of the 188 worked solutions; for example EE 2026 Q.62, EC 2023 Q.14 and EC 2022 Q.32.
• Using a vector without scaling it to unit length — Taking a dot product with a direction vector, or choosing a normal vector, without dividing by its length. The result is then off by the length of that vector, or the chosen vector is not a unit vector. Recorded in at least 5 of the 188 worked solutions; for example EC 2026 Q.12, EE 2025 Q.53 and EE 2024 Q.36.
• Deciding the number of solutions without comparing ranks — Concluding how many solutions a linear system has from the determinant, or from the count of equations and unknowns, without comparing the rank of the coefficient matrix with the rank of the augmented matrix. Recorded in at least 4 of the 188 worked solutions; for example IN 2026 Q.14, EE 2025 Q.13 and EC 2022 Q.12.
• Dropping a chain-rule or product-rule term — Differentiating a composite or product expression and leaving out the extra factor or term that the chain rule or product rule supplies, which shifts the derivative and the final answer. Recorded in at least 4 of the 188 worked solutions; for example EE 2026 Q.50, EC 2020 Q.3 and EE 2018 Q.30.

Networks (Electric Circuits) (EC, EE and IN)

• Misreading polarity or reference direction of a source or current — Reading a battery, dependent source or port current with the opposite polarity or direction to the one drawn. The sign of a term in the equation flips and the result has the wrong sign or magnitude. Recorded in at least 18 of the 122 worked solutions; for example EE 2026 Q.24, EC 2024 Q.31 and IN 2025 Q.45.
• Wrong units: rad/s for Hz, degrees for radians, milli for micro — Entering a quantity in the wrong unit or scale, such as angular frequency for frequency, degrees for radians or mF for microfarads. The numbers then differ by a factor of 2π, a power of ten, or a trigonometric value. Recorded in at least 12 of the 122 worked solutions; for example EC 2026 Q.51, IN 2026 Q.33 and EE 2023 Q.51.
• Treating peak values as rms, or the reverse — Using an amplitude where an rms value is needed, or the other way round, scales the answer by a factor of root two or two. The power, current or energy then comes out wrong by exactly that factor. Recorded in at least 11 of the 122 worked solutions; for example EC 2026 Q.31, IN 2025 Q.60 and EE 2026 Q.42.
• Opening a voltage source or shorting a current source — When finding a Thevenin resistance or a time constant, sources are switched off the wrong way: a voltage source opened or a current source shorted. This leaves the wrong resistors in the network. Recorded in at least 10 of the 122 worked solutions; for example EE 2026 Q.60, IN 2026 Q.39 and EC 2024 Q.32.
• Mixing up line and phase quantities in three-phase circuits — Applying the star relations to a delta load, or putting a phase voltage where the line voltage belongs, and the reverse. The result is off by a factor of root three. Recorded in at least 7 of the 122 worked solutions; for example EE 2024 Q.50, IN 2022 Q.55 and EE 2023 Q.39.
• Using series-resonance rules for a parallel circuit, or the reverse — A series resonance condition or Q formula is applied to a circuit with a parallel branch, or a parallel formula is applied to a series circuit. The two cases give different conditions and different Q, so the answer fails. Recorded in at least 6 of the 122 worked solutions; for example IN 2026 Q.16, EC 2023 Q.18 and EE 2018 Q.55.
• Using the wrong starting value of the capacitor or inductor — Assuming a capacitor starts uncharged or at the full source voltage, or ignoring the initial current, instead of finding the true pre-switching value. The whole transient then starts from the wrong point. Recorded in at least 6 of the 122 worked solutions; for example EE 2026 Q.43, IN 2025 Q.64 and EE 2025 Q.57.
• Treating a quantity that can jump as continuous at switching — Only the capacitor voltage and inductor current are continuous across a switching instant. The slip is carrying over a resistor voltage, node voltage or capacitor current from before the switch, which gives the wrong value just after it. Recorded in at least 5 of the 122 worked solutions; for example IN 2025 Q.63, EC 2023 Q.35 and EE 2023 Q.29.

Signals and Systems (EC, EE and IN)

• Dropping a constant scale factor such as 1/N, 1/(2π) or one half — Leaving out the normalising factor in a transform pair, inverse DFT, Parseval relation or modulation property. The shape of the answer is right but its size is off by N, 2π or 2. Recorded in at least 13 of the 118 worked solutions; for example EE 2025 Q.61, EC 2024 Q.48 and IN 2024 Q.53.
• Judging linearity or time-invariance from how the formula looks — Calling a system nonlinear or time-varying from a visible factor, or running the time-invariance test with the delay applied in the wrong place. The verdict then follows the appearance of the formula and not the definition. Recorded in at least 7 of the 118 worked solutions; for example EE 2026 Q.14, IN 2026 Q.20 and EC 2023 Q.26.
• Reading the region of convergence wrongly — Taking the ROC from time limits, from the wrong side of a pole, or as a union of half-planes. The causality, stability or existence conclusion then fails. Recorded in at least 7 of the 118 worked solutions; for example EC 2026 Q.16, EE 2024 Q.16 and IN 2024 Q.38.
• Shifting or reflecting in the wrong direction — Moving a signal, spectrum or impulse the wrong way, or mixing up delay and advance. The breakpoints, index or exponent land on the wrong side. Recorded in at least 7 of the 118 worked solutions; for example IN 2026 Q.41, EE 2025 Q.20 and IN 2018 Q.33.
• Confusing compression with stretching in time scaling — Treating a time-compressed signal as longer, or multiplying by the scale factor where it should divide. Periods, energies and bandwidths move the wrong way. Recorded in at least 6 of the 118 worked solutions; for example EC 2026 Q.43, IN 2022 Q.18 and EE 2024 Q.53.
• Finding a period by the wrong shortcut — Using the continuous-time period formula on a discrete signal, taking the LCM of frequency coefficients instead of periods, or reading the period off a single term. The result is not the fundamental period. Recorded in at least 6 of the 118 worked solutions; for example EE 2026 Q.12, IN 2026 Q.19 and EC 2024 Q.56.
• Confusing rad/s with Hz — Reading an angular-frequency value as hertz, or dividing by π instead of 2π. Bandwidths and cutoffs then come out off by a factor of 2π. Recorded in at least 5 of the 118 worked solutions; for example EC 2026 Q.46, IN 2025 Q.41 and EE 2025 Q.61.
• Not squaring amplitudes when finding energy or power — Adding amplitudes or samples directly, or leaving a scale factor unsquared, where energy or power needs squares. The total or the scaling then comes out wrong. Recorded in at least 5 of the 118 worked solutions; for example EC 2026 Q.43, EE 2025 Q.30 and IN 2024 Q.13.

Control Systems (EC, EE and IN)

• Building the characteristic equation from the wrong loop terms — Working from the open-loop denominator, dropping a zero or feedback term, or ignoring an inner loop. The polynomial is then not the closed-loop one, so the stability or damping result is wrong. Recorded in at least 7 of the 119 worked solutions; for example EE 2025 Q.41, EC 2020 Q.23 and IN 2018 Q.38.
• Reading ω² as ω, or taking an arithmetic mean for a geometric one — Reading the squared quantity as the frequency itself, or averaging corner frequencies arithmetically. The frequency is then wrong by a square root. Recorded in at least 7 of the 119 worked solutions; for example EC 2023 Q.25, EE 2023 Q.38 and IN 2023 Q.43.
• Applying the final value theorem without checking stability — The theorem only gives the true steady-state value when the closed loop is stable. Using it on an unstable loop returns a finite number that the response does not settle to. Recorded in at least 6 of the 119 worked solutions; for example IN 2026 Q.59, EC 2024 Q.13 and EE 2018 Q.25.
• Misreading loops, branch directions or signs in a diagram — Using the standard negative-feedback formula where the diagram differs, treating a feedback arc as forward, or counting a minus sign once instead of everywhere it applies. The transfer function then has the wrong loops or signs. Recorded in at least 6 of the 119 worked solutions; for example EC 2026 Q.19, EE 2024 Q.20 and IN 2022 Q.37.
• Sign slips in the determinant or adjugate of sI − A — Writing the wrong sign on an off-diagonal entry of the matrix or its adjugate. The characteristic polynomial or transfer function then has the wrong coefficients, and the poles or numerator come out wrong. Recorded in at least 6 of the 119 worked solutions; for example EC 2026 Q.41, EE 2026 Q.52 and IN 2026 Q.44.
• Taking an angle in the wrong quadrant or axis — Reading an angle from a calculator arctangent when the vector lies in the second quadrant, or measuring it from the wrong real axis. The phase or damping angle is then off by one hundred and eighty degrees or by the wrong complement. Recorded in at least 6 of the 119 worked solutions; for example IN 2025 Q.47, EE 2024 Q.57 and IN 2024 Q.45.
• Mixing up the direction or sign convention of Nyquist encirclements — Using the encirclement rule for the wrong contour direction, or swapping P minus Z with Z minus P. The count of right-half-plane closed-loop poles then comes out with the wrong sign. Recorded in at least 5 of the 119 worked solutions; for example EE 2025 Q.23, IN 2025 Q.28 and EC 2020 Q.11.
• Converting decibels with the power formula or the wrong direction — Using ten times the log instead of twenty times, or inverting the ratio, when turning a decibel value into a gain. The resulting gain is wrong by a large factor. Recorded in at least 4 of the 119 worked solutions; for example IN 2026 Q.60, EE 2024 Q.59 and EC 2018 Q.42.

Analog Electronics (EC, EE and IN)

• Reading a battery, diode or Zener polarity backwards — Taking the orientation of a battery symbol, diode or Zener the wrong way round. The conduction state, clamp level or voltage polarity is then reversed. Recorded in at least 10 of the 112 worked solutions; for example EC 2026 Q.65, IN 2024 Q.41 and EE 2023 Q.37.
• Losing the sign of an inverting stage — Dropping the minus sign of an inverting amplifier or integrator, or the inversion at a transistor drain. The magnitude is right but the answer has the wrong polarity. Recorded in at least 9 of the 112 worked solutions; for example EC 2025 Q.21, IN 2024 Q.59 and EE 2022 Q.16.
• Mixing up inverting and non-inverting gain formulas — Using the inverting-amplifier relation for a stage where the signal enters the plus input, or the reverse, or dropping the 1 + term. The gain then comes out with the wrong magnitude or sign. Recorded in at least 8 of the 112 worked solutions; for example EC 2026 Q.17, IN 2024 Q.56 and EE 2026 Q.37.
• Applying the virtual short where it does not hold — Treating the op-amp input as a virtual short or virtual ground without checking for negative feedback, a grounded input or linear operation. The assumed node voltage is then wrong and so is everything computed from it. Recorded in at least 7 of the 112 worked solutions; for example IN 2025 Q.20, EC 2020 Q.8 and EE 2022 Q.16.
• Misidentifying the amplifier configuration — Classifying a MOSFET or BJT stage from a remembered pattern instead of from where the signal enters and leaves. A common-gate stage is read as inverting, high-input-resistance or a follower, and the gain, sign or impedance is wrong. Recorded in at least 7 of the 112 worked solutions; for example IN 2026 Q.21, EC 2025 Q.30 and EE 2025 Q.16.
• Unit and scale slips: Hz, rad/s, kHz, milli, percent — Giving the right number in the wrong unit or off by a power of ten, such as rad/s for Hz, ohms for milliohms or percent for a per-degree figure. The value is correct in size but fails the unit the question asks for. Recorded in at least 7 of the 112 worked solutions; for example IN 2026 Q.61, EE 2020 Q.53 and EC 2020 Q.41.
• Ignoring base current or the (β + 1) factor in bias loops — Treating a bias divider as stiff, leaving out the base current, or scaling a resistor by β instead of β + 1. The base-loop equation is then wrong and the collector current or node voltage is off. Recorded in at least 6 of the 112 worked solutions; for example EC 2026 Q.18, IN 2026 Q.22 and EE 2024 Q.42.
• Leaving out the base-emitter drop — Equating a node voltage to the Zener or divider voltage without subtracting the base-emitter drop of the transistor. The emitter or output level, and every quantity derived from it, is then off. Recorded in at least 4 of the 112 worked solutions; for example EE 2026 Q.33, IN 2025 Q.35 and EC 2020 Q.35.

Digital Electronics (EC, EE and IN)

• Overlooking an inversion bubble or inverter in the figure — Missing a small bubble or inverter on a gate or wire, or reading a bubbled gate as a plain one, flips the logic of that stage and sends the whole circuit to a wrong function or address. Recorded in at least 8 of the 103 worked solutions; for example IN 2025 Q.42, EC 2020 Q.10 and IN 2023 Q.42.
• Using a flip-flop's value after the clock edge, not before it — In a synchronous circuit every flip-flop reads the old state at the edge. The slip is updating stages one at a time, or reading a signal just after the edge, so a changed value is fed into the next stage too early. Recorded in at least 7 of the 103 worked solutions; for example IN 2026 Q.47, EC 2025 Q.45 and IN 2024 Q.42.
• Applying De Morgan's theorem halfway or missing its equal forms — Complementing only some of the literals or operators, applying the theorem to gates it does not concern, or not seeing that two options are the same function after De Morgan leads to a wrong or incomplete answer. Recorded in at least 5 of the 103 worked solutions; for example EE 2026 Q.45, IN 2022 Q.23 and IN 2025 Q.29.
• Reading a Q-bar connection as Q, or the reverse — Taking a feedback or gate input from Q when the figure draws it from the complemented output (or the other way round) changes the next-state equations, so the state sequence or frequency comes out wrong. Recorded in at least 5 of the 103 worked solutions; for example EC 2026 Q.49, IN 2024 Q.42 and EE 2023 Q.56.
• Counting one transition or gate delay as a full period — An output needs two transitions, one up and one down, to complete a cycle. Treating a single delay, toggle or state change as the period gives a frequency that is too high. Recorded in at least 4 of the 103 worked solutions; for example EC 2026 Q.35, IN 2023 Q.41 and EC 2024 Q.41.
• Dropping the 1.76 dB term or mixing 10 log and 20 log in ADC noise — Quoting converter signal-to-noise ratio with only the rough 6 dB per bit, or applying the amplitude log to a power ratio, shifts the answer away from the correct value. Recorded in at least 4 of the 103 worked solutions; for example IN 2026 Q.23, EC 2025 Q.44 and EC 2024 Q.46.
• Leaving out the don't-care cells in a K-map — Treating the don't-care minterms as 0 breaks the groups into smaller ones, so the simplified expression is longer than the one in the options. Recorded in at least 3 of the 103 worked solutions; for example EC 2024 Q.40, IN 2020 Q.8 and EE 2018 Q.37.
• Putting the wrong variable or bit order on a multiplexer's select lines — The select inputs of a multiplexer are assigned to the wrong bit positions — the least significant select line taken as the most significant — or a variable that is not a select input is treated as one. Each data input then lands on the wrong minterm and the function comes out wrong. Recorded in at least 3 of the 103 worked solutions; for example EC 2026 Q.48, EE 2024 Q.26 and EC 2020 Q.10.

Electronic Devices (EC)

• Getting a sign or direction backwards — Reversing a sign or a direction, such as forward-bias depletion width, the field against diffusion, the sign of the flat-band charge term or a band-diagram arrow, so the result moves the wrong way. Recorded in at least 6 of the 37 worked solutions; for example EC 2024 Q.26, EC 2024 Q.61 and EC 2020 Q.6.
• Unit conversion slips that are out by a power of ten — Mixing up length, area, diffusivity or resistivity units, such as nanometres or micrometres to centimetres, square metres per second to square centimetres per second, or ohm-metres to kilo-ohm-metres, so the answer is out by a power of ten. Recorded in at least 5 of the 37 worked solutions; for example EC 2026 Q.64, EC 2024 Q.61 and EC 2020 Q.34.
• Using a formula or operating region where it does not apply — Applying a standard formula or assuming a device stays in one region of operation without checking the condition that makes it valid, such as saturation versus triode or a junction with no donors. Recorded in at least 5 of the 37 worked solutions; for example EC 2024 Q.52, EC 2024 Q.64 and EC 2018 Q.45.
• Swapping electron and hole quantities or donor and acceptor roles — Using the hole value where the electron value belongs, inverting a mobility ratio, or reversing which dopant type supplies which carrier. Recorded in at least 4 of the 37 worked solutions; for example EC 2025 Q.31, EC 2023 Q.56 and EC 2023 Q.16.

Electromagnetics (EC, EE and IN)

• Carrying over a result from a different geometry or setting — A familiar formula for a cylinder, point charge, waveguide or fixed-voltage case is applied where the problem has another geometry or condition, giving the wrong dependence or value. Recorded in at least 9 of the 71 worked solutions; for example IN 2026 Q.15, EC 2023 Q.20 and EE 2023 Q.61.
• Losing a minus sign or reversing the order of a vector product — The slip is dropping the sign in a field or potential relation, or taking the cross product in the wrong order, which flips the direction of the result and points to a mirror-image answer. Recorded in at least 7 of the 71 worked solutions; for example EE 2026 Q.16, IN 2026 Q.15 and IN 2026 Q.38.
• Mixing up radius and diameter or leaving lengths unconverted — Substituting a diameter for a radius, or leaving centimetres, inches or metres unconverted, scales the answer by a wrong power of two or ten. Recorded in at least 6 of the 71 worked solutions; for example EC 2026 Q.21, IN 2022 Q.59 and EC 2026 Q.33.
• Using Cartesian derivatives or area elements in curved coordinates — In cylindrical or spherical coordinates the divergence, curl and area or volume elements carry extra factors of r or sin θ. Leaving them out gives an expression that cannot vanish or integrates to the wrong value. Recorded in at least 5 of the 71 worked solutions; for example EE 2025 Q.51, EE 2024 Q.52 and EE 2022 Q.23.
• Confusing series and parallel combinations — Stacked dielectric layers, angular sectors or resistors in a loop are treated as the wrong kind of combination, series for parallel or the reverse. The equivalent capacitance or resistance then comes out wrong. Recorded in at least 4 of the 71 worked solutions; for example EC 2025 Q.65, EE 2025 Q.50 and IN 2024 Q.15.
• Mixing field-amplitude and power ratios — Converting decibels or reflection coefficients as if a power quantity were an amplitude, or the reverse, applies the square or the square root the wrong number of times. Recorded in at least 4 of the 71 worked solutions; for example EC 2026 Q.21, EC 2026 Q.34 and EC 2018 Q.33.
• Treating a varying density or field as uniform — Using a single value of a quantity that changes with position, instead of integrating it, gives a wrong total charge, current or flux. Recorded in at least 3 of the 71 worked solutions; for example EC 2025 Q.49, EE 2022 Q.23 and EE 2020 Q.55.
• Using the free-space wavelength inside a medium or guide — The condition depends on the wavelength in the coating, glass or guide, not in free space. Using the free-space value gives a wrong thickness or dimension. Recorded in at least 3 of the 71 worked solutions; for example EC 2023 Q.60, IN 2020 Q.47 and EC 2018 Q.48.

Electrical Machines (EE and IN)

• Applying the turns ratio once or the wrong way — Applying the turns ratio once where it is needed twice, in the wrong direction, or not at all when the question asks for a value on a specific side. The result is off by a factor of the ratio or its square. Recorded in at least 4 of the 50 worked solutions; for example IN 2026 Q.17, EE 2024 Q.54 and IN 2026 Q.58.
• Getting leading versus lagging or a phase-shift sign backwards — Reversing the sign for a leading or lagging power factor, a reactive drop or a delta-star phase shift. The error flips the answer to the opposite option. Recorded in at least 4 of the 50 worked solutions; for example EE 2026 Q.22, EE 2025 Q.43 and EE 2022 Q.64.
• Holding a variable fixed that the machine couples to another — Assuming speed, current, flux or slip stays constant while another quantity changes, when the machine or load couples them. The ratio calculation then loses a factor. Recorded in at least 4 of the 50 worked solutions; for example EE 2025 Q.44, EE 2023 Q.48 and EE 2022 Q.62.
• Scaling a quantity linearly when it varies as a square — Treating copper loss, eddy loss or transformed current as proportional to the first power of the ratio, when it follows the square. The answer is off by a square factor. Recorded in at least 4 of the 50 worked solutions; for example EE 2026 Q.31, IN 2026 Q.17 and EE 2022 Q.65.
• Mixing up poles and pole pairs in conversions — Using poles where pole pairs belong, or the reverse, in synchronous speed and in mechanical-to-electrical angle. The error halves or doubles the answer. Recorded in at least 3 of the 50 worked solutions; for example EE 2025 Q.35, IN 2022 Q.29 and EE 2022 Q.35.
• Skipping field current or loss between line and armature values — Moving between line current and armature current without removing the field current, or between shaft output and input without adding the loss. The current or power is then misstated. Recorded in at least 3 of the 50 worked solutions; for example EE 2024 Q.55, EE 2022 Q.64 and EE 2020 Q.28.

Electrical and Electronic Measurements (EE and IN)

• Combining component errors by the wrong rule — Errors in several components are combined with a rule that does not suit the question, such as subtracting them, using root-sum-square for a worst-case limit, adding percentages where absolute errors are needed, or the reverse. The stated error then comes out wrong. Recorded in at least 7 of the 43 worked solutions; for example IN 2026 Q.29, EE 2025 Q.31 and IN 2023 Q.64.
• Pairing adjacent bridge arms instead of opposite ones — A bridge balance condition multiplies opposite arms. Pairing arms that sit next to each other gives a wrong or impossible balance relation. Recorded in at least 6 of the 43 worked solutions; for example EE 2026 Q.57, IN 2025 Q.56 and IN 2024 Q.39.
• Wrong unit or power-of-ten prefix in the answer — A prefix such as pico, nano, kilo or milli is taken as the wrong power of ten, or the final figure is quoted in volts, henry or hertz when the question asks for millivolts, millihenry or kilohertz. The answer is then off by a factor of ten or more. Recorded in at least 6 of the 43 worked solutions; for example EE 2025 Q.59, IN 2020 Q.43 and IN 2025 Q.56.
• Treating series elements as parallel, or the reverse — Reading how two or three elements are connected the wrong way round gives a wrong equivalent resistance or Q, and so a wrong current or reading. Recorded in at least 5 of the 43 worked solutions; for example IN 2025 Q.26, IN 2023 Q.30 and IN 2022 Q.63.
• Taking rad/s as Hz, or leaving out the 2π — Treating an angular frequency as a frequency in hertz, or stopping at ω without converting, puts a factor of 2π into the wrong place. Recorded in at least 4 of the 43 worked solutions; for example IN 2025 Q.56, IN 2018 Q.20 and IN 2024 Q.39.
• Using a small-change or equal-arm shortcut outside its range — A shortcut that holds only for equal arms, tiny changes, or voltages far above the diode's thermal voltage is applied where the data do not meet that condition. The result is far from the correct value. Recorded in at least 3 of the 43 worked solutions; for example EE 2025 Q.59, IN 2022 Q.65 and EE 2020 Q.52.

Programming in C (CS)

• Being off by one when tracing a loop — Stopping the trace one pass early, running it one pass too long, or miscounting how many times a loop halves or repeats, so the final count or value is one step out. Recorded in at least 5 of the 34 worked solutions; for example CS 2025 Set 1 Q.63, CS 2025 Set 2 Q.19 and CS 2025 Set 2 Q.33.
• Mixing up scope, lifetime and storage of variables — Confusing where a variable is visible with how long it lives, for example treating a static local as heap memory, reusing an inner variable outside its block, or resetting a static variable on each call. Recorded in at least 5 of the 34 worked solutions; for example CS 2026 Set 1 Q.34, CS 2020 Q.46 and CS 2026 Set 2 Q.17.
• Losing track of what a pointer currently refers to — Assuming a pointer still refers to its first target, or that a pointer parameter behaves like a reference, so the wrong variable is read or changed. Recorded in at least 4 of the 34 worked solutions; for example CS 2026 Set 2 Q.60, CS 2018 Q.29 and CS 2025 Set 1 Q.34.
• Losing track of what each nested or recursive call contributes — Counting a nested or recursive call once when it is applied at every level, or flattening the levels of recursion, so the result is built from the wrong number of calls. Recorded in at least 3 of the 34 worked solutions; for example CS 2026 Set 1 Q.61, CS 2025 Set 2 Q.63 and CS 2023 Q.36.

Data Structures and Algorithms (CS)

• Overlooking a special structure that allows a cheaper algorithm — Assuming the general, costlier method is needed when the problem's structure (acyclic graph, tree only, heap, weaker requirement) permits a faster one. The complexity is then overstated. Recorded in at least 5 of the 84 worked solutions; for example CS 2026 Set 2 Q.37, CS 2026 Set 2 Q.38 and CS 2025 Set 2 Q.20.
• Misclassifying edge types in DFS and BFS traversals — Assuming an edge type can or cannot occur in a given traversal, or reversing the nesting that defines it. Statements about back, forward and cross edges are then judged wrongly. Recorded in at least 4 of the 84 worked solutions; for example CS 2026 Set 1 Q.50, CS 2025 Set 2 Q.29 and CS 2024 Set 1 Q.45.
• Not carrying a complexity derivation through to the end — Stopping partway in a recurrence or loop analysis, such as skipping a substitution or leaving a sum unevaluated. The final order of growth is then off by a factor. Recorded in at least 4 of the 84 worked solutions; for example CS 2026 Set 1 Q.17, CS 2025 Set 1 Q.20 and CS 2024 Set 1 Q.42.
• Off-by-one in heap indexing and the leaf boundary — Using the wrong index base or the wrong rounding for the split between internal nodes and leaves. Parent-child positions shift by one and the leaf count is wrong. Recorded in at least 4 of the 84 worked solutions; for example CS 2026 Set 1 Q.23, CS 2024 Set 1 Q.43 and CS 2023 Q.12.
• Expecting a singly linked list to allow fast backward access — Carrying array habits over to a linked list, such as binary search or an O(1) step back. A singly linked list has no random access and no back pointer, so these steps cost a linear scan. Recorded in at least 3 of the 84 worked solutions; for example CS 2023 Q.13, CS 2020 Q.16 and CS 2018 Q.3.
• Mishandling the probe sequence in open addressing — Wrongly numbering, wrapping or recomputing probes when a slot is occupied. Keys land in the wrong slot and later answers shift. Recorded in at least 3 of the 84 worked solutions; for example CS 2026 Set 1 Q.24, CS 2025 Set 1 Q.65 and CS 2020 Q.23.
• Reading height as a count of nodes or levels — Giving height or path length as a number of nodes or levels when it is defined in edges. The answer is one more than the correct value. Recorded in at least 3 of the 84 worked solutions; for example CS 2025 Set 1 Q.35, CS 2025 Set 2 Q.35 and CS 2018 Q.20.

Computer Organization and Architecture (CS)

• Forgetting associativity when counting cache sets and index bits — Dividing cache size by block size and stopping, without dividing by the number of ways. The set count, index width and tag width are then wrong. Recorded in at least 4 of the 57 worked solutions; for example CS 2026 Set 1 Q.38, CS 2023 Q.64 and CS 2020 Q.30.
• Ignoring opcode space already taken by other fields or formats — Counting opcode patterns as if the whole field were free, without removing bits used by addressing modes, type bits or shorter opcodes of other formats. The count of usable opcodes is too high. Recorded in at least 4 of the 57 worked solutions; for example CS 2026 Set 2 Q.44, CS 2024 Set 2 Q.57 and CS 2020 Q.44.
• Mishandling bits versus bytes and size conventions — Mixing bits with bytes, decimal with binary sizes, or element size with address step. The figures come out off by a factor of 8, a power of two or the element width. Recorded in at least 4 of the 57 worked solutions; for example CS 2024 Set 1 Q.54, CS 2024 Set 2 Q.11 and CS 2023 Q.41.
• Misjudging how much CPU work an I/O scheme leaves — Crediting a scheme with less CPU involvement, or better utilisation, than it provides, or mixing a hardware scheme with its software look-alike. The ranking of schemes is then reversed. Recorded in at least 4 of the 57 worked solutions; for example CS 2026 Set 2 Q.18, CS 2024 Set 1 Q.15 and CS 2022 Q.17.
• Off by one cycle, bit or word in a count — Leaving out an initial bit, a fill or drain cycle, or counting total cycles where only stall cycles belong. The answer is off by one unit. Recorded in at least 4 of the 57 worked solutions; for example CS 2026 Set 1 Q.60, CS 2025 Set 2 Q.32 and CS 2020 Q.21.
• Applying the wrong cycle-time or cycle-count formula to a pipeline — Using total latency, a sum of stage delays or a non-pipelined count where the slowest stage or the fill-and-drain formula applies. Execution time comes out wrong. Recorded in at least 3 of the 57 worked solutions; for example CS 2025 Set 2 Q.56, CS 2023 Q.33 and CS 2018 Q.50.
• Charging a penalty to all accesses, not the affected share — Charging a miss or stall cost to all instructions or accesses rather than only the share that triggers it, or without conditioning on the level above. The average cost is overstated. Recorded in at least 3 of the 57 worked solutions; for example CS 2025 Set 2 Q.55, CS 2022 Q.61 and CS 2024 Set 1 Q.56.

Fundamentals of Biological Engineering (BT)

• Substituting numbers without converting units first — Entering quantities given in different units straight into a formula, such as mg with g, minutes with seconds or cp with Pa s. The result is wrong by the missing conversion factor. Recorded in at least 9 of the 85 worked solutions; for example BT 2026 Q.32, BT 2026 Q.55 and BT 2026 Q.61.
• Using a base-10 logarithm where the natural log belongs — Taking a base-10 logarithm where the natural logarithm is needed, because rates such as growth, death and heat-transfer averages are defined through an exponential. The resulting time or value is then wrong. Recorded in at least 8 of the 85 worked solutions; for example BT 2026 Q.59, BT 2025 Q.28 and BT 2024 Q.35.
• Counting electrons per molecule instead of per carbon atom — Reporting the degree of reduction or available electrons for the whole molecule rather than per carbon atom. Missing the division by the carbon count gives a value several times too large. Recorded in at least 5 of the 85 worked solutions; for example BT 2025 Q.26, BT 2025 Q.58 and BT 2024 Q.33.
• Using a yield coefficient inverted or on the wrong basis — Multiplying where the yield divides, or mixing up mass, molar and carbon-mole bases for a yield. The amount comes out wrong. Recorded in at least 4 of the 85 worked solutions; for example BT 2025 Q.58, BT 2023 Q.48 and BT 2022 Q.28.
• Confusing the specific growth rate with the doubling time — Treating a rate as a time, or the reciprocal of a time as the growth rate, without the natural log of 2 that links the two. The answer lands on the wrong quantity or is off by that factor. Recorded in at least 3 of the 85 worked solutions; for example BT 2025 Q.28, BT 2022 Q.33 and BT 2022 Q.65.
• Leaving out the starting amount already present — Computing only the newly formed biomass or cell count and dropping what was there at the start. The total then comes out too small. Recorded in at least 3 of the 85 worked solutions; for example BT 2024 Q.58, BT 2022 Q.58 and BT 2022 Q.28.
• Treating dilution rate and flow rate as the same quantity — Using the dilution rate in place of the volumetric flow rate, or inverting it, in a chemostat problem. Flow rate is the dilution rate times the volume, so the swap gives a wrong flow rate or a false washout. Recorded in at least 3 of the 85 worked solutions; for example BT 2026 Q.61, BT 2025 Q.62 and BT 2020 Q.49.

Genetics, Cell and Molecular Biology (BT)

• Swapping two look-alike terms with different definitions — Taking one technical term for its near neighbour, such as totipotent for pluripotent or apoptosis for necrosis. The definition of the other term decides the answer. Recorded in at least 7 of the 72 worked solutions; for example BT 2024 Q.16, BT 2024 Q.19 and BT 2024 Q.43.
• Swapping enzymes that have similar-sounding jobs — Picking a polymerase, repair enzyme or replication factor that sounds right but works on a different template, product or step. The mismatch in template or function is what rules it out. Recorded in at least 6 of the 72 worked solutions; for example BT 2026 Q.22, BT 2025 Q.40 and BT 2025 Q.52.
• Using the wrong cross or reference group in genetics probability — Using the wrong parental genotypes, expected ratio or reference group, or letting earlier offspring change the odds. The probability then belongs to a different question. Recorded in at least 5 of the 72 worked solutions; for example BT 2026 Q.64, BT 2023 Q.30 and BT 2025 Q.30.
• Assuming there is only one mechanism, route or possibility — Treating the best-known mechanism as the only one, such as a single reading frame or a single way to silence a gene. The overlooked alternative is what the question tests. Recorded in at least 4 of the 72 worked solutions; for example BT 2026 Q.42, BT 2020 Q.29 and BT 2020 Q.3.
• Dropping a conversion or doubling step in nucleic acid arithmetic — Skipping the triplet-codon factor, a unit conversion or the pairing step when moving between bases, amino acids, lengths and percentages. The result is off by a fixed factor. Recorded in at least 4 of the 72 worked solutions; for example BT 2025 Q.31, BT 2019 Q.48 and BT 2023 Q.55.
• Stopping at the familiar step and missing another — Marking the best-known correct item or step and overlooking another that the question also requires. The answer is then incomplete. Recorded in at least 4 of the 72 worked solutions; for example BT 2026 Q.43, BT 2024 Q.51 and BT 2022 Q.43.
• Putting an event or chromosome count in the wrong phase — Assigning an event to the wrong phase or division, or counting chromosomes and kinetochores for the wrong stage. The phase decides the count or the order. Recorded in at least 3 of the 72 worked solutions; for example BT 2025 Q.50, BT 2020 Q.48 and BT 2019 Q.6.

Recombinant DNA Technology and Tools (BT)

• Skipping a unit or angular-frequency conversion — Carrying values in mM, mol per L, mg per mL or Hz into a formula that needs other units. The result is off by a power of ten or by 2π. Recorded in at least 5 of the 53 worked solutions; for example BT 2026 Q.35, BT 2025 Q.64 and BT 2023 Q.65.
• Choosing an option because the name sounds right — Picking an option on surface wording, such as a familiar keyword or a name that suggests thoroughness, rather than the property the question tests. The actual property rules it out. Recorded in at least 4 of the 53 worked solutions; for example BT 2023 Q.26, BT 2023 Q.45 and BT 2019 Q.8.
• Confusing one detection technique with another that resembles it — Mixing up blots, microarrays and fragment-based markers, or detection of a modification with measuring expression. The molecule each technique targets decides the answer. Recorded in at least 4 of the 53 worked solutions; for example BT 2025 Q.55, BT 2023 Q.24 and BT 2022 Q.26.
• Mixing up expression hosts and transformation methods — Matching a host or method to the wrong category, such as treating a bacterial strain as a cell line or a vector-based method as direct. The defining property of the host or method rules the option out. Recorded in at least 4 of the 53 worked solutions; for example BT 2025 Q.20, BT 2024 Q.22 and BT 2023 Q.44.
• Swapping enzymes that do opposite or unrelated jobs — Choosing an enzyme that sounds suitable but does the opposite job or acts on a different substrate. Phosphatase for kinase and digestive enzyme for reporter enzyme are examples. Recorded in at least 4 of the 53 worked solutions; for example BT 2023 Q.22, BT 2020 Q.37 and BT 2019 Q.27.
• Applying a rule without checking its condition — Using a counting rule, enzyme step or separation outcome without checking the condition it needs, such as a circular versus linear molecule or a reducing agent. The rule fails where its condition does not hold. Recorded in at least 3 of the 53 worked solutions; for example BT 2022 Q.64, BT 2020 Q.37 and BT 2020 Q.52.
• Crediting a structural technique with what it does not measure — Assigning a spectroscopic or imaging method the wrong kind of information, or the wrong hardware. Each method reports a specific property, so the mismatch fails the question. Recorded in at least 3 of the 53 worked solutions; for example BT 2024 Q.41, BT 2023 Q.28 and BT 2019 Q.32.
• Getting a direction or strand orientation the wrong way round — Reversing which way a charge, strand or flow of information runs. A primer copied the way it is printed or a charge sent to the wrong electrode gives the opposite result. Recorded in at least 3 of the 53 worked solutions; for example BT 2020 Q.27, BT 2020 Q.7 and BT 2019 Q.33.

Every example above opens the question’s own page, with the official answer, the worked solution and the common-mistake note in full. The same questions can be seen in the order each paper printed them on the paper pages, and each subject’s notes and formula sheet are in the study library.