Measurements and Theory of Errors - Revision Notes
Surveying is fundamentally the science of measured distances, angles, and elevations — but every measurement contains error. The PRC Civil Engineer board exam consistently tests error classification, most probable value (MPV), tape corrections, and error propagation. This chapter provides the statistical and physical framework that underpins all subsequent surveying topics: leveling, traversing, and curve layout. Mastery here is non-negotiable for licensure candidates.
Classification of Errors
Exam tips
- Board exam MCQs often ask: 'Which type of error can be corrected by applying a formula?' — Answer: Systematic.
- If a question says 'the tape was too long' or 'temperature was higher than standard,' it is describing a systematic error.
- Watch for trick questions that describe a blunder (e.g., 'surveyor read 6 m instead of 9 m') and ask for its statistical treatment — the answer is: none; it must be discarded.
Key points
- Three classes: Mistakes (blunders), Systematic errors, and Random (accidental) errors.
- Mistakes are gross human blunders — misreading a tape, wrong booking — and must be detected and eliminated, not statistically treated.
- Systematic errors follow a definite physical law (e.g., thermal expansion, elastic stretch) and are correctable by formula.
- Random errors are small, unavoidable, and follow the normal (Gaussian) distribution — these are treated with probability theory.
- Only random errors are subject to statistical analysis (MPV, probable error, error propagation).
- Accuracy refers to closeness to the true value; Precision refers to repeatability/consistency of measurements.
Definitions
Mistake (Blunder)
A gross error due to carelessness or misunderstanding, such as misreading a tape or transposing digits.
Importance
Must be identified and removed before statistical analysis. Cannot be corrected by formula.
Systematic Error
An error that follows a definite mathematical or physical law and always acts in the same direction under the same conditions.
Importance
Tape temperature error, sag error, and tension error are all systematic — they are computed and applied as corrections.
Random (Accidental) Error
Small, residual errors remaining after blunders and systematic errors are removed; governed by the laws of probability.
Importance
The basis for computing MPV, probable error, and confidence intervals in repeated measurements.
Accuracy vs. Precision
Accuracy = closeness of a measurement to the true value. Precision = closeness of repeated measurements to each other (repeatability).
Importance
A survey instrument can be precise but inaccurate (systematic bias). Both must be controlled for quality work.
Common mistakes
- Applying statistical formulas (MPV, probable error) to systematic errors — only random errors are treated statistically.
- Confusing accuracy with precision — high precision does not guarantee high accuracy.
- Attempting to 'average out' a systematic error instead of computing and applying the physical correction.
- Forgetting that mistakes must be eliminated (re-measured), not mathematically corrected.
Most Probable Value and Probable Error
Formulas
MPV = x̄ = Σx / n
Example
Measurements: 100.1, 100.2, 100.0 m → MPV = (100.1 + 100.2 + 100.0)/3 = 300.3/3 = 100.1 m
Variables
x̄ = most probable value; Σx = sum of all measurements; n = number of measurements
Application
Used whenever a quantity is measured multiple times under the same conditions with equal precision.
E = 0.6745 √(Σv² / (n−1))
Example
For MPV = 100.1 m: v₁ = 0.0, v₂ = +0.1, v₃ = −0.1 m; Σv² = 0.02 m²; E = 0.6745√(0.02/2) = 0.6745×0.1 = ±0.0674 m
Variables
E = probable error of a single observation; v = residual = (x_i − x̄); n = number of observations
Application
Quantifies the precision of a single measurement from a set of repeated observations.
E_m = E / √n
Example
From above, E_m = 0.0674/√3 = ±0.0389 m. The mean 100.1 m has a smaller error than any single reading.
Variables
E_m = probable error of the mean; E = probable error of a single observation; n = number of observations
Application
Shows how the mean becomes more reliable as more measurements are taken.
Exam tips
- Board problem pattern: Given n measurements, find MPV and E_m. Always set up a residuals table: x_i, v_i = x_i − x̄, v_i².
- If the problem gives you E of a single observation and asks for E_m, simply divide E by √n.
- Memorize 0.6745 — this constant is non-negotiable and is not given in the exam.
- A question asking 'what is the precision of the mean after 4 observations if one observation has E = ±0.04 m?' → E_m = 0.04/√4 = ±0.02 m.
Key points
- The Most Probable Value (MPV) of a set of equally weighted, repeated measurements is the arithmetic mean.
- Residuals (v) are the differences between each individual measurement and the MPV: v = x_i - x_bar.
- The Probable Error of a single observation E = 0.6745 × sqrt(Σv² / (n−1)).
- The Probable Error of the mean E_m = E / sqrt(n) — taking more measurements reduces the error of the mean.
- The constant 0.6745 comes from the normal distribution: it is the value z such that P(|error| ≤ z·σ) = 0.50.
- Probable error means there is a 50% chance the true error lies within ±E of the measured value.
- For unequal weights, the weighted mean is used: x_bar = Σ(w_i · x_i) / Σw_i, where weights are often inversely proportional to variance.
Definitions
Residual (v)
The difference between an individual measurement and the mean: v = x_i − x̄. Not to be confused with the true error (x_i − true value).
Importance
Residuals are computable from the data; true errors require knowledge of the exact true value.
Probable Error (E)
A statistical measure such that there is a 50% probability that the true random error of a single observation does not exceed ±E.
Importance
Standard way of expressing precision in classical surveying; appears frequently in Philippine board exam problems.
Probable Error of the Mean (E_m)
The probable error applicable to the mean (MPV) of n observations; always smaller than E by factor 1/√n.
Importance
Demonstrates why more measurements improve reliability — a key justification for repeated observations in geodetic surveys.
Common mistakes
- Using n instead of (n−1) in the denominator of the variance formula — (n−1) is the correct divisor for sample standard deviation.
- Computing residuals as (x̄ − x_i) instead of (x_i − x̄) — the sign does not matter for v² but keep it consistent.
- Confusing E (probable error of one shot) with E_m (probable error of the mean) — the mean is always more reliable.
- Forgetting to take the square root of (Σv²/(n−1)) before multiplying by 0.6745.
Tape (Distance) Corrections
Formulas
C_t = α · L · (T − T_s)
Example
50 m tape, α = 11.6×10⁻⁶/°C, T = 40°C, T_s = 20°C: C_t = 11.6×10⁻⁶ × 50 × 20 = +0.0116 m per tape length.
Variables
α = coefficient of thermal expansion (steel ≈ 11.6×10⁻⁶ /°C); L = measured length (m); T = field temperature (°C); T_s = standard temperature (°C)
Application
Applied whenever the field temperature differs from the tape's standardization temperature.
C_p = (P − P_s) · L / (A · E)
Example
P = 80 N, P_s = 50 N, L = 30 m, A = 6.45 mm² = 6.45×10⁻⁶ m², E = 200×10⁹ Pa: C_p = (30)(30)/(6.45×10⁻⁶ × 200×10⁹) = 900/1,290,000 = +0.000698 m ≈ +0.70 mm
Variables
P = applied tension (N); P_s = standard tension (N); L = measured length (m); A = cross-sectional area of tape (m²); E = modulus of elasticity of steel ≈ 200 GPa
Application
Corrects for elastic stretch when the tape is pulled at a tension different from its standardization tension.
C_sag = −w²L³ / (24P²) or equivalently −W²L / (24P²)
Example
w = 0.04 N/m, L = 30 m, P = 50 N: C_sag = −(0.04²×30³)/(24×50²) = −(0.0016×27000)/60000 = −43.2/60000 = −0.00072 m
Variables
w = weight of tape per unit length (N/m); W = total weight of tape between supports (N); L = unsupported length (m); P = tension (N)
Application
Applies when the tape is supported only at its ends (not throughout). Sag is always a shortening of the measured distance.
C_h = −h² / (2L) (approximate) or H = √(L² − h²) (exact)
Example
L = 100 m, h = 4 m: H = √(100² − 4²) = √9984 = 99.92 m. Check via approx: C_h = −16/200 = −0.08 m → H ≈ 99.92 m. ✓
Variables
h = vertical difference in elevation between tape ends (m); L = slope distance (m); H = horizontal distance (m)
Application
C_h is the approximate slope correction for gentle slopes (h/L ≤ 0.10). Use exact formula for steep terrain.
True Length = Measured Length × (Actual Tape Length / Nominal Tape Length)
Example
30 m tape actually 30.02 m; measured distance reads 150 m: True = 150 × (30.02/30) = 150 × 1.000667 = 150.10 m
Variables
Actual tape length = true physical length of the tape (measured by comparison); Nominal tape length = marked/stated length of tape
Application
Used when the tape is found to be longer or shorter than its nominal length (e.g., worn, stretched, or improperly fabricated).
Exam tips
- Board exam frequently combines two or three corrections in one problem. Tabulate each correction separately, then sum algebraically.
- Quick check for tape-too-long direction: 'A long tape gives short readings; a short tape gives long readings' — think of it as each step being bigger or smaller than marked.
- For temperature problems, remember that steel tapes are standardized at 20°C in the Philippines. If field T = 40°C, C_t is positive (tape is longer → measured reading is shorter → true is longer).
- When asked for horizontal distance from slope data, always show both exact and approximate formulas — the approximate C_h = −h²/(2L) is sufficient for gentle slopes and is faster in the exam.
- Sag correction: W is total weight of tape = w × L. Double-check which variable form the problem gives you before substituting.
Key points
- Four main systematic corrections for steel tape measurements: Temperature (C_t), Tension/Pull (C_p), Sag (C_sag), and Slope (C_h).
- Corrections are added algebraically: True Length = Measured Length + ΣCorrections.
- Temperature correction C_t = α·L·(T − T_s): positive when T > T_s (tape longer, measured distance shorter → true is longer).
- Tension correction C_p = (P − P_s)·L / (A·E): positive when applied tension P > standard P_s.
- Sag correction C_sag = −w²L³/(24P²) or −W²L/(24P²): ALWAYS NEGATIVE — sagging tape is shorter between support points.
- Slope correction: use exact formula H = √(L² − h²), or approximate C_h = −h²/(2L) for gentle slopes (h/L < 0.1).
- Tape too long: each 'full tape length' laid down is actually longer than nominal → true measured distance > reading → multiply by (actual/nominal).
- Tape too short: true measured distance < reading → multiply by (actual/nominal), which is now < 1.
- When LAYING OUT (setting a distance): reverse the correction — divide by (actual/nominal) or subtract what you added when measuring.
- Standard conditions for a Philippine tape: P_s = 50 N (sometimes 5 kg), T_s = 20°C, tape supported throughout (no sag in standard condition).
Definitions
Standard Conditions
The specific temperature (T_s, usually 20°C), tension (P_s, usually 50 N), and support conditions under which a tape is calibrated.
Importance
All corrections are relative to standard conditions — if field conditions equal standard conditions, no correction is needed.
Normal (Pull-Out) Tension
The tension at which the elongation due to pull exactly cancels the shortening due to sag — the tape reads true length with end support only.
Importance
Appears in advanced tape problems; computed from C_p + C_sag = 0.
Absolute vs. Comparative Measurement
Absolute: measuring against a known standard unit. Comparative: measuring relative to another object (e.g., comparing tape to a reference tape).
Importance
Calibration of tapes is a comparative measurement — the actual tape length is found by comparison to a standard.
Common mistakes
- CRITICAL: Reversing the sign when tape is too long vs. too short — if tape is LONG, each measured span covers MORE ground → true > measured → add correction.
- CRITICAL: Reversing correction direction when LAYING OUT vs. MEASURING — when laying out, the correction is reversed.
- Sag correction sign: students sometimes forget C_sag is always negative. There is no situation where sag makes a tape read longer.
- Using total weight W in place of weight per unit length w in the sag formula — check which form of the formula you are using.
- Applying slope correction C_h with a positive sign — the horizontal distance is ALWAYS less than the slope distance, so C_h is always negative.
- Using E (modulus of elasticity) from the tension correction formula in the probable error formula — these are completely different quantities with the same letter.
- Forgetting that if multiple corrections apply, all must be applied cumulatively: True = Measured + C_t + C_p + C_sag + C_h.
Error Propagation
Formulas
E_sum = √(E₁² + E₂² + ... + E_n²)
Example
Three distances measured with errors ±0.03, ±0.04, ±0.05 m: E_sum = √(0.03² + 0.04² + 0.05²) = √(0.0009+0.0016+0.0025) = √0.005 = ±0.0707 m
Variables
E_sum = probable error of the sum; E₁, E₂, ..., E_n = probable errors of individual measurements
Application
Used when a total distance or quantity is the sum of independently measured parts with different or equal errors.
E_sum = E · √n (for n equal-error measurements)
Example
A traverse has 5 lines, each measured with E = ±0.02 m: Error of total traverse = 0.02√5 = ±0.0447 m
Variables
E = probable error of each individual measurement; n = number of measurements summed
Application
Used when the same measurement (same error) is repeated n times and results are added (e.g., n tape lengths).
E_A = √[(W · E_L)² + (L · E_W)²]
Example
L = 50 m (E_L = ±0.05 m), W = 30 m (E_W = ±0.03 m): E_A = √[(30×0.05)² + (50×0.03)²] = √[2.25 + 2.25] = √4.5 = ±2.12 m²
Variables
E_A = error in computed area; E_L = error in length L; E_W = error in width W
Application
Error propagation in a rectangular area where both dimensions are independently measured.
Exam tips
- A very common board problem: 'A line is measured n times with probable error ±E each time. Find the probable error of the TOTAL length.' Answer: E√n.
- Another common form: 'Find the probable error of the MEAN.' Answer: E/√n. Know which one is being asked.
- Error propagation for sums: the total error is always LESS than the arithmetic sum of individual errors but GREATER than the largest single error.
- For area problems, always write out E_A = √[(W·E_L)² + (L·E_W)²] and substitute carefully — units must be consistent.
Key points
- When a final result depends on several independently measured quantities, errors in each component propagate into the final result.
- For a sum or difference of measured quantities: E_total = √(E₁² + E₂² + E₃² + ...).
- For a series of n equal-precision measurements summed: E_sum = E_single × √n.
- This shows why errors grow with more measurements but more slowly than linearly — errors partially cancel.
- For a product or quotient, relative errors (E/x) propagate: (E_result/result)² = (E₁/x₁)² + (E₂/x₂)².
- Error propagation in an area (A = L × W): E_A = √[(W·E_L)² + (L·E_W)²].
- The relationship E_sum = E√n vs. E_mean = E/√n is a key exam pairing — sum grows, mean shrinks.
Definitions
Error Propagation
The process of computing the uncertainty in a derived quantity (sum, product, function) given the uncertainties in the directly measured quantities.
Importance
Critical for assessing total survey accuracy — a traverse with 10 courses has more total error than one with 4 courses, but not proportionally.
Independent Measurements
Measurements whose errors are uncorrelated — an error in one does not influence the others.
Importance
The quadratic (root-sum-of-squares) propagation formula assumes independence. Correlated errors require more complex treatment.
Common mistakes
- Adding errors arithmetically (E_sum = E₁ + E₂ + ...) instead of in quadrature (root-sum-of-squares) — arithmetic addition always overestimates total error.
- Confusing E_sum = E√n (sum grows) with E_mean = E/√n (mean shrinks) — these are both correct but apply to different questions.
- Forgetting to multiply partial derivatives when propagating errors through products — e.g., for A = L × W, the error in A is not just √(E_L² + E_W²).
- Applying error propagation to systematic errors — only random errors propagate statistically.
Connections
- Error propagation (E_sum = E√n) directly governs the allowable misclosure in closed traverses — Philippine surveying practice sets permissible closure limits based on this principle.
- Tape corrections (temperature, tension, sag, slope) are preconditions for accurate distance measurement in plane table surveying, stadia work, and EDM calibration checks.
- The MPV concept extends to angle measurement in theodolite work — repetition method and direction method both rely on averaging to reduce random error.
- Probable error and standard error concepts connect directly to leveling: allowable error in spirit leveling is expressed as E = C√K, where K is distance in km — the same √n relationship.
- RA 544 (Civil Engineering Law of the Philippines) mandates that licensed engineers adhere to standards of survey accuracy — understanding error theory is a legal professional requirement, not merely academic.
- Error propagation in areas connects to land registration under the Land Registration Authority (LRA) and cadastral surveys governed by DENR administrative orders.
- The normal tension concept in tape corrections is a special case of elastic mechanics — the same Young's modulus E (200 GPa for steel) appears in structural steel design under AISC 360 and NSCP 2015 structural provisions.
- In GPS/GNSS surveying (modern geomatics), dilution of precision (DOP) is essentially an error propagation coefficient — the classical theory of this chapter is the conceptual ancestor of GPS accuracy metrics.
Quick review questions
A steel tape of 30 m nominal length is found to be 30.015 m long. A distance is measured as 450 m. What is the true length of the line?
Answer
True length = 450 × (30.015/30) = 450 × 1.0005 = 450.225 m
The tape is TOO LONG — each span it lays down covers more ground than the nominal 30 m reads. True = Measured × (Actual/Nominal). Since actual > nominal, true > measured. Always verify the direction: a long tape gives short readings, so the true distance must be longer.
A 50 m tape (α = 11.6 × 10⁻⁶/°C, standardized at 20°C) is used to measure a 300 m distance at a field temperature of 38°C. What is the temperature correction?
Answer
C_t = α·L·(T − T_s) = 11.6×10⁻⁶ × 300 × (38−20) = 11.6×10⁻⁶ × 300 × 18 = +0.0626 m
Temperature is above standard, so the tape has expanded → it is longer → each span covers more ground than marked → the true distance is longer than the measured distance → C_t is positive. True length = 300 + 0.0626 = 300.063 m.
Five measurements of a horizontal angle give residuals of +2″, −3″, +1″, +4″, −4″. Compute the probable error of a single observation.
Answer
Σv² = 4+9+1+16+16 = 46 (arc-seconds²); E = 0.6745√(46/(5−1)) = 0.6745√11.5 = 0.6745×3.391 = ±2.29″
Steps: (1) List residuals. (2) Square each: 4, 9, 1, 16, 16. (3) Sum: Σv² = 46. (4) Divide by (n−1) = 4: 11.5. (5) Take square root: 3.391. (6) Multiply by 0.6745: ±2.29 arc-seconds. This is E, the probable error of one observation.
From the previous problem, what is the probable error of the mean angle?
Answer
E_m = E/√n = 2.29/√5 = 2.29/2.236 = ±1.02″
The mean of 5 observations is more reliable than any single observation. E_m = E/√n. The probable error of the mean is always smaller than that of a single measurement. This is why field procedures specify a minimum number of repetitions.
A slope distance of 250 m is measured between two points with an elevation difference of 10 m. Find the horizontal distance using (a) the exact formula and (b) the approximate slope correction.
Answer
(a) H = √(250² − 10²) = √(62500 − 100) = √62400 = 249.80 m. (b) C_h = −h²/(2L) = −100/500 = −0.20 m → H ≈ 249.80 m.
Both methods agree to mm here since h/L = 10/250 = 0.04 (gentle slope). The approximate formula C_h = −h²/(2L) is derived from the binomial expansion of √(L²−h²) and is valid for gentle slopes (h/L < 0.10). For steeper slopes, always use the exact formula.
A surveyor measures a traverse leg 6 times with a probable error of ±0.03 m each time. (a) What is the probable error of the total distance if all 6 measurements are summed? (b) What is the probable error of the mean?
Answer
(a) E_sum = 0.03√6 = 0.03 × 2.449 = ±0.0735 m. (b) E_mean = 0.03/√6 = ±0.01225 m.
These two results address different questions: (a) the total accumulated error if 6 legs are laid end-to-end (sum grows as √n), versus (b) the reliability of the average of 6 repeated measurements of the same leg (mean improves as 1/√n). Both use √n but in opposite ways.
A tape is supported only at its ends. Weight of tape = 0.048 N/m, unsupported length = 30 m, applied tension = 60 N. Compute the sag correction.
Answer
C_sag = −w²L³/(24P²) = −(0.048²×30³)/(24×60²) = −(0.002304 × 27000)/(86400) = −62.208/86400 = −0.000720 m ≈ −0.72 mm
The sag correction is ALWAYS negative — a sagging tape is geometrically shorter between its support points than a straight tape. Check: C_sag must be negative. If you get a positive answer, a sign error has occurred. The true distance is longer than the tape reading by |C_sag|.
A rectangular lot is measured as L = 60 m ± 0.06 m and W = 40 m ± 0.04 m. Compute the probable error of the computed area.
Answer
E_A = √[(W·E_L)² + (L·E_W)²] = √[(40×0.06)² + (60×0.04)²] = √[(2.4)² + (2.4)²] = √(5.76+5.76) = √11.52 = ±3.39 m²
Area errors propagate through partial derivatives: ∂A/∂L = W, ∂A/∂W = L. The formula E_A = √[(W·E_L)² + (L·E_W)²] follows directly. Both terms contributed equally (±2.4) in this symmetric case. Area = 60×40 = 2,400 m² ± 3.39 m² (relative precision ≈ 1 in 709).
Diagrams
Measurements and Theory of Errors — Chapter Overview
A hierarchical mind map showing all major topics of the chapter: error classification, most probable value and probable error formulas, tape correction types, and error propagation rules. Use this as a quick-reference overview before starting detailed review.
Decision Flowchart: Applying Tape Corrections
A step-by-step decision flowchart guiding the user through identifying which tape corrections apply, computing each correction with the correct sign, and combining them to find the true length. Highlights the critical sign rules: sag and slope corrections are always negative; the measuring vs. laying-out distinction for tape-length errors.
Step-by-Step Procedure: Computing MPV and Probable Error
An 8-step procedural flowchart for computing MPV, probable error of a single observation, and then branching to compute either the probable error of the mean (E/√n) or the probable error of a sum (E√n). Helps students avoid the most common board exam mistake of confusing these two quantities.
Distribution of Error Theory Topics in PRC Board Exam Questions
A pie chart showing the approximate distribution of board exam questions from this chapter based on historical PRC exam patterns. Tape corrections dominate at ~45%, followed by MPV and probable error calculations at ~25%. Prioritize tape corrections when allocating study time.
Error Propagation: Sum vs. Mean — Key Decision Tree
A decision tree clarifying the three main error propagation scenarios: (1) summing distances uses E·√n or root-sum-of-squares, (2) taking the mean uses E/√n, and (3) computing an area uses the partial-derivative-based formula. Highlights the critical conceptual distinction between sum errors growing and mean errors shrinking as n increases.
Exam strategy
In the PRC board exam, Surveying typically comprises 15–20% of the Mathematics, Surveying, and Transportation Engineering (MSTE) paper. For this chapter: (1) ALWAYS classify the error type first — systematic corrections use formulas; random errors use statistics. (2) For tape correction problems, list all corrections, compute each separately, and sum algebraically — a tabular approach avoids sign errors. (3) Remember the two golden mnemonics: 'LONG tape → SHORT reading → add to get true' and 'SAGS always subtract.' (4) For MPV/probable error, always set up a residuals table before computing. (5) In propagation problems, distinguish between E_sum = E√n (total of a series) vs. E_mean = E/√n (average) — exam writers deliberately test this distinction. (6) Time allocation: tape correction problems with 2–3 corrections take about 3–4 minutes each; MPV/probable error problems take 4–5 minutes. Allocate accordingly.