Reading Stamina for Long English Exams | Keep Accuracy After the Easy Questions

51. Build a stamina profile by failure signature

Do not label the learner simply “low stamina”. Record what actually changes late: source mixing, broad rereading, longer answers, missing components, lower confidence, faster guessing, weaker summaries, mechanical transfer errors or reduced checking quality. The failure signature determines the training target. Two learners can both decline after fifty minutes for entirely different reasons.

52. Profile the time of first drift

Identify roughly when the process first begins to change. It may be minute thirty-five, after the second passage, or only during the final source set. The exact minute is less important than the transition. Stamina training should begin near that boundary so the learner practises preserving quality where it currently becomes unstable.

53. Profile the first mechanism to drift

Accuracy often falls after process has already changed. The first drift may be weaker passage mapping, poorer source labels, more answer alternatives, less local evidence search or less complete response planning. Train that earliest change before waiting for visible mark loss. Endurance repair is stronger when it catches the upstream deterioration.

54. Profile whether the drift is gradual or sudden

A gradual decline suggests cumulative load. A sudden drop after one difficult question or passage suggests a recovery event or confidence shock. The same late score can therefore imply different intervention. Leonie distinguishes slope from cliff: slow erosion needs endurance architecture; abrupt collapse may need recovery first.

55. Profile whether the drift is universal or conditional

If late accuracy falls across all genres and question types, general endurance may be weak. If it falls only on paired sources, multimodal tasks or summary, the stamina problem is conditional. Train the vulnerable combination late rather than extending every type of task equally.

56. Profile whether the learner becomes slower or faster

Some students slow under fatigue because they reread and overcheck. Others speed up because they want to finish. Both can lose accuracy. The stamina profile should record direction of timing drift. The repair for overcautious slowing is different from the repair for impulsive acceleration.

57. Late slowness often reflects rising uncertainty cost

As confidence falls, the learner may spend longer proving answers that were already evidence-sufficient. Checking becomes reassurance. Train objective stop criteria under late conditions: command met, evidence sufficient, scope matched, answer complete. When these are satisfied, uncertainty can remain without reopening the task.

58. Late speed often reflects collapsing standards

A tired learner may stop checking quantifiers, skip answer planning or select the first plausible option. Completion gets faster but less reliable. Protect a minimum decision floor: read the command, identify the evidence job, perform one verification where risk is high. Endurance requires standards that do not disappear when urgency rises.

59. Build a stable-zone benchmark

Measure what good execution looks like while fresh. How concise are one-mark answers? How often are sources misattributed? How local is evidence search? What level of certainty language is used? This early benchmark becomes the standard the learner tries to preserve later. Stamina is relative to the learner’s own stable process.

60. Build a late-zone benchmark

Run a realistic longer session and capture the same variables in the final third. The difference between early and late states becomes the endurance gap. Training aims to narrow that gap rather than chase vague feelings of mental toughness.

61. Use paired early–late items

Place comparable question families early and late in a session. Two inference items of similar difficulty, two source comparisons, two summary-selection tasks. The learner’s performance difference helps isolate position effects. This is more useful than comparing completely different tasks at different times.

62. Keep difficulty controlled during early stamina training

If the late task is much harder, failure can be blamed on complexity rather than endurance. Begin with ordinary material placed later. Once accuracy is preserved, increase difficulty. Training should first establish that the learner can carry known competence through duration.

63. Use duration ladders

Move from the current stable duration to slightly longer windows rather than jumping directly to the full examination every time. Forty minutes stable may lead to fifty, then sixty. Each step is accepted only if the target controls survive. Duration is increased as a training variable, not endured as punishment.

64. Use late-position ladders

A skill can be moved progressively later without lengthening the entire session dramatically. Comparison may be practised at minute twenty, then thirty-five, then fifty. This isolates endurance of the mechanism while keeping the rest of the session manageable.

65. Use load ladders

After duration stabilises, add cognitive load: more sources, subtler distinctions, denser syntax, more switching. Increase one dimension at a time. A learner who can preserve source identity for fifty minutes on simple texts may not yet preserve it for fifty minutes on three-source synthesis. Load and duration interact.

66. Use interruption-free practice selectively

Long uninterrupted sessions are necessary at times because real examinations require sustained continuity. They should not dominate every week. Component repair still benefits from shorter high-quality work. Full continuity is a specific endurance condition, not the default format for all learning.

67. Use passage-boundary resets consistently

The reset should become automatic: close previous source, preserve any flag, check time state, identify the new passage job, begin. Consistency matters because the reset is most needed when the learner is tired enough to forget it. A routine that depends on remembering to be strategic has not fully automated.

68. Reset source identity at each new text

Late source blending often happens because ownership from earlier texts stays mentally active. Starting a new source should explicitly clear the old voice and define the new one. In multi-source tasks, create the source map before reasoning. This prevents cumulative identity clutter.

69. Reset confidence at each new question

A difficult previous item should not determine how certain the next answer feels. Confidence must be rebuilt from the new evidence. This reset protects the learner from both panic underconfidence and compensatory overconfidence. Each question earns its own certainty level.

70. Reset reading resolution at each new passage

One dense argument may require close attention; the next narrative may not. Carrying the previous passage’s slow, defensive reading mode forward wastes stamina. The learner should adapt resolution to the new text rather than remain trapped in the previous difficulty state.

71. Working-memory conservation begins before the final third

Endurance is easier when early tasks do not consume unnecessary memory. Externalise source labels, close completed questions, keep annotations functional and stop generating extra interpretations. Saving cognitive resources early is part of late accuracy even though the benefit appears later.

72. Answer inflation is a stamina tax

Every unnecessary sentence costs writing time, decision time and later checking time. Over a long paper, small excesses accumulate. Concise complete responses protect the final third by preventing earlier questions from consuming disproportionate resources.

73. Over-annotation is a stamina tax

Marking everything creates visual clutter and adds processing. The late learner then has to navigate both the passage and a dense layer of annotations. Keep only marks that reduce future retrieval or preserve state. Stamina improves when the external system is lighter, not merely when the student tries harder.

74. Repeated option comparison is a stamina tax

On multiple-choice items, a learner may repeatedly cycle through the same two options. Require new evidence for another cycle. Without new evidence, preserve the stronger supported choice and move. Repetitive comparison consumes time and attention that later questions need.

75. Unresolved ambiguity is a stamina tax when left unparked

An uncertain item can stay active in working memory after the learner moves. A brief breadcrumb externalises the unresolved state and reduces mental rehearsal. The return becomes cheaper and the rest of the paper is protected from one unanswered question.

76. Excessive internal narration is a stamina tax

Some learners silently explain every decision in full sentences. Self-explanation is useful during learning but expensive during performance. Compress it to the decisive rule: “final stance”, “Source B”, “condition preserved”, “two reasons”. Automaticity should replace long metacognitive commentary.

77. Stable answer contracts reduce late decision load

When the learner recognises the structure of a comparison, inference or effect question automatically, fewer resources are spent deciding how to format the response. This frees attention for evidence and meaning. Familiar logical architecture is an endurance asset.

78. Stable evidence corridors reduce late search load

Late broad rereading is expensive. A reliable passage map allows the learner to predict a region even when memory feels weaker. Training should repeatedly connect question job to passage location so local retrieval survives duration.

79. Stable certainty language reduces late scope drift

Students should have a small functional range: states, shows, suggests, may indicate, strongly implies. The aim is not a phrase bank but a calibrated ladder. When tired, the learner can select from a familiar evidence-strength system rather than inventing certainty language each time.

80. Stable comparison axes reduce source-switch fatigue

Once the shared dimension is known, each source can be read in relation to it. Without the axis, the learner repeatedly switches and reinterprets. Late comparison becomes especially expensive. Train axis-first thinking until it remains available under duration.

81. Late summary work needs protected selection architecture

When tired, learners are tempted to draft immediately. Protect the two-stage process: select distinct in-category propositions, then compress. The selection phase can be brief but should not disappear. Otherwise fatigue creates duplicates, category errors and expensive rewrites.

82. Late multimodal work needs a mode-job reset

A tired learner may respond to the most visually salient source. Before interpreting, name the mode’s job: chart for trend, image for instance, prose for explanation, caption for context. This simple reset prevents late salience from replacing evidential relevance.

83. Late evaluation needs criterion discipline

Fatigue encourages generic judgements such as “reliable because expert”. Protect one step: what claim am I evaluating, and which criterion matters most? The criterion narrows the reasoning and reduces cognitive load. Claim-sensitive evaluation is both more accurate and more efficient late.

84. Late tone questions need degree control

When tired, subtle distinctions collapse into extreme labels. Use broad dimensions first: positive/negative, certain/cautious, warm/distant. Then choose degree. This reduces lexical search and protects against overstatement in the final third.

85. Late purpose questions need scale control

A learner under load may answer every purpose question globally because whole-text purpose is easier to remember than local function. Check the requested scale before reasoning. Local sentence, paragraph and whole passage require different evidence. Scale recognition is a low-cost endurance control.

86. Late structure questions need movement control

Feature listing becomes tempting when mental resources are low. Return to before → move → after. This simple architecture preserves the relational answer even when detailed terminology is harder to retrieve. Movement is more important than decorative feature inventory.

87. Late direct-detail questions should remain simple

After many difficult questions, the learner may distrust an easy direct item and overthink it. Reset to the command. If one explicit proposition answers it, use that proposition and move. Endurance includes preserving simplicity where the task is simple.

88. Late inference should not become story generation

As working memory declines, learners may rely more on intuitive narrative rather than evidence. Re-anchor to the smallest supported bridge. Evidence → inference → calibrated wording. The compact sequence protects against imagination filling gaps that the text did not earn.

89. Late checking should become more selective

110 owns the final checking system. Under fatigue, the key endurance question is whether the learner can still prioritise certain loss, incomplete structures and high-risk flags rather than rereading everything. Selectivity is itself stamina because it preserves the remaining cognitive budget.

90. Use a small late-paper risk card in training

The card might contain three items: source, scope, complete. Another learner may need map, evidence, stop. Keep it short. The card is a temporary external support for the specific late failure signature. Fade it when the routine becomes automatic.

91. Train fatigue-resistant source ownership

Place paired-source tasks late in sessions. Require explicit source labels and one shared axis. Reduce support gradually. The success criterion is not merely correct comparison while fresh, but correct ownership after cumulative load.

92. Train fatigue-resistant scope

Use late-session near-miss answers that differ only by quantifier, certainty, condition or time frame. The learner must select the wording the evidence earns. These short drills build proposition fidelity under load without requiring another full mock.

93. Train fatigue-resistant answer completeness

Place two-part questions late and require the learner to state the answer contract before writing. Over time, fade the explicit statement. The aim is for “two reasons”, “both sources” and “device + effect” to remain visible even when working memory is tired.

94. Train fatigue-resistant evidence search

Late questions should still begin with target prediction and local search. If the learner broadens prematurely, reduce passage length or complexity until local retrieval survives. Then rebuild load. Endurance training preserves search discipline rather than accepting late rereading as inevitable.

95. Train fatigue-resistant uncertainty tolerance

Use late items where one defensible answer remains moderately uncertain. The learner must release it once objective criteria are met. This protects the paper from late reassurance loops. Endurance includes the ability to finish without perfect subjective certainty.

96. Train fatigue-resistant recovery

Place one controlled hard question before a late ordinary passage. The target is not solving the hard item; it is restoring normal execution afterward. Compare answer length, rereads, confidence and source control before and after the disruption. Recovery quality is part of stamina when the disruption occurs early enough to affect the remaining duration.

97. Train fatigue-resistant switching

Late in a session, switch from prose to graph, narrative to argument or single-source to paired-source. Use micro-resets so the learner carries the question state without carrying the wrong discourse frame. Switching endurance is not the same as simply working longer.

98. Train fatigue-resistant mechanics

Answer transfer, flag management, scrolling, numbering and submission must remain accurate at the end. Practise these mechanical actions under realistic duration. Intellectual comprehension can be strong and still lose marks through late workflow failure.

99. The endurance system should taper explicit effort

Early training may feel deliberate because many controls are consciously monitored. Mature stamina should feel simpler. Maps are lighter, source identity is automatic, answer contracts are familiar and resets are brief. The system survives long duration because it demands less conscious management.

100. The final-third standard is ordinary excellence, not heroic effort

The learner should not need a dramatic motivational surge to survive the final passage. The goal is ordinary execution under accumulated load: read the new text, build a usable model, find evidence, answer the job, preserve scope, move. Endurance is mature when the final third looks recognisably like the first third, even if it feels harder.

101. Practice Lab: Maren is accurate early and incomplete late

Maren’s early answers satisfy every component. In the final third, two-part questions increasingly receive one part. The passage understanding remains sound. Her endurance failure sits in answer construction. Training moves short multi-part questions to the late phase of sessions and requires a compact contract check before writing. The goal is not deeper reading but preserving completeness after duration.

102. Practice Lab: Iona’s sources blend only after fifty minutes

Iona compares sources accurately while fresh. Late paired-source work shows correct ideas assigned to the wrong speaker. The skill exists but loses source ownership under load. She begins using explicit late source labels and comparison-axis notes. The same task is practised at minute twenty and minute fifty until the gap narrows.

103. Practice Lab: Leonie becomes slower, not less accurate

Leonie preserves marks late but uses increasing time because she rereads and overchecks. The visible score looks stable, yet completion risk rises. Her endurance problem is uncertainty cost. Training uses objective stop criteria late in sessions. She learns that moderate subjective doubt can coexist with an evidence-sufficient answer.

104. Practice Lab: Maren becomes faster and less accurate

Maren reacts to fatigue by accelerating. In the final third she stops checking question quantity, drops conditions and chooses plausible options quickly. The repair protects a minimum decision floor: command, evidence, scope, complete. She is allowed to remain fast only while those controls survive.

105. Practice Lab: time debt is mistaken for stamina failure

Iona’s late answers are rushed. Review shows she arrives six minutes behind because the first passage was over-annotated. When early annotation is reduced, late accuracy normalises without longer endurance sessions. The symptom appeared late, but the cause sat early. 120 and 030 repair the upstream inefficiency; 140 does not overclaim ownership.

106. Practice Lab: genuine stamina failure remains after pacing is fixed

Leonie now reaches the final third on schedule, yet source labels and quantifiers still degrade. The timing hypothesis has been cleared. Duration itself, or the accumulated cognitive load within it, remains relevant. Late-position source and scope drills enter the programme.

107. Practice Lab: a hard early question poisons the whole paper

Maren spends several minutes on an ambiguous inference and eventually skips it. She technically recovers, but later answers become long and defensive. The issue is not ordinary duration. The confidence shock remains active. Recovery training now includes restoration: after parking the hard item, the next simple question must return to normal answer length and evidence standards.

108. Practice Lab: the same duration is easy after a smooth opening

Iona can sustain eighty minutes accurately when early passages are ordinary, but deteriorates after an unusually difficult opening. This conditional endurance pattern shows that cumulative uncertainty, not clock duration alone, drives the decline. Training includes controlled early disruption followed by normal work.

109. Practice Lab: the map disappears late

Leonie maps the first passage carefully and then stops mapping later because it feels like extra effort. Evidence search broadens. Her repair is not to restore a full early map. She uses a lean late map: paragraph job, turn, source, conclusion. Retrieval improves with little added load.

110. Practice Lab: the map stays but becomes too detailed

Maren responds to fatigue by writing more notes because she trusts memory less. The extra annotation itself becomes burdensome. Training limits the map to navigation value. Confidence is transferred from volume of notes to reliability of a few structural anchors.

111. Practice Lab: late vocabulary uncertainty freezes the passage

Iona can ignore nonessential unfamiliar terms early. Late in the exam, the same kind of uncertainty triggers repeated rereading. Her lexical triage routine is practised at the end of longer sessions: essential or nonessential, contextual cue, decide, move. The goal is preserving uncertainty tolerance.

112. Practice Lab: dense syntax feels harder only late

Leonie parses complex sentences accurately at minute fifteen but loses conditions at minute sixty. Syntax knowledge is present. The endurance boundary is local parsing under load. Short dense-sentence tasks are moved late in realistic sessions until the main-proposition → condition → qualifier routine remains stable.

113. Practice Lab: the final direct question is overthought

After difficult source work, Maren reaches a simple direct-detail question. She distrusts the obvious answer and spends two minutes seeking hidden meaning. A passage-boundary reset and command check teach her to treat the new question on its own terms. Endurance includes preserving simplicity after complexity.

114. Practice Lab: late inference becomes a story

Iona’s early inferences are tightly bounded. Late ones add motives and future consequences not supported by evidence. Her late risk card contains one cue: smallest bridge. At minute fifty, she must still state the clue and the minimum supported conclusion before writing.

115. Practice Lab: late tone becomes too extreme

Leonie distinguishes cautious, sceptical and critical early. Late, everything negative becomes hostile. Training reduces the late choice to dimensions and degree. She identifies direction first, then intensity. The simpler decision architecture preserves nuance under fatigue.

116. Practice Lab: late purpose becomes generic

Maren answers local-function questions with broad statements such as “to persuade the reader”. Her whole-text model is still available, so it overrides the smaller task. A late scale check—sentence, paragraph or whole text—restores the correct search size and answer precision.

117. Practice Lab: late structure becomes a feature list

Iona can explain structural movement while fresh. Late she names “anecdote, statistic, short paragraph” without relation. The stamina routine reduces structure to before → move → after. Once movement is stated, only useful features are attached.

118. Practice Lab: late comparison loses symmetry

Leonie gives Source A a full interpretation and Source B only a quotation. Her answer is not a reading failure; it is a late completeness failure in a cognitively expensive family. Training uses a visible A | axis | B skeleton late in sessions, then fades it.

119. Practice Lab: late evaluation defaults to authority

Maren’s early evaluation is claim-sensitive. Late she writes “reliable because expert”. The endurance control is criterion first. What claim is being evaluated, and which criterion governs it? The one-step reset narrows the reasoning enough to survive fatigue.

120. Practice Lab: late summary begins drafting too early

Iona reaches the summary tired and starts writing immediately. Duplicates and category errors follow. Training protects a short selection stage even at the end: mark distinct in-category propositions, then compress. The phase is brief, but it is never skipped.

121. Practice Lab: late summary overcompresses

Leonie removes conditions and actor identity to save words and finish faster. Her late summary training focuses on semantic minimums: actor, relation, condition where necessary. Concision cannot cross the boundary where proposition identity disappears.

122. Practice Lab: graph labels blur late

Maren reads charts accurately early but confuses groups late. A final-third visual routine protects variable, unit and group before trend interpretation. This is a mechanical micro-check with high endurance value because visual salience becomes less trustworthy under load.

123. Practice Lab: image salience wins late

Iona knows that a photograph illustrates one instance, yet late in a multimodal set she treats it as proof of prevalence. The mode-job reset—instance, magnitude, cause, context—re-establishes evidential roles. Fatigue is prevented from turning vividness into relevance.

124. Practice Lab: scrolling creates late broad rereads

Leonie’s paper-based stamina is strong. Digitally, she scrolls past evidence and repeatedly restarts. The endurance limit is partly interface load. Training uses digital landmarks, flags and consistent return points. Reading stamina is repaired through mechanical fluency, not just longer reading.

125. Practice Lab: answer-sheet transfer fails only at the end

Maren understands the paper but late transfer errors appear when attention is low. The mechanical task is integrated into the endurance plan with a protected transfer checkpoint and final alignment scan. Comprehension performance includes the delivery chain.

126. Practice Lab: handwriting causes under-answering

Iona’s late two-mark answers become one short phrase because writing feels costly. She practises concise complete late responses rather than longer handwriting drills. The target is minimum scoring structure, not physical output volume.

127. Practice Lab: typing creates late perfectionism

Leonie repeatedly edits wording because digital text is easy to modify. This creates time loss and new errors. A stop rule—clear, complete, scope-correct—limits editing. Endurance is protected by reducing optional late revision.

128. Practice Lab: full mocks reproduce the same late collapse

Maren sits several ninety-minute mocks and always loses source control after hour one. More mocks provide the same measurement. Training shifts to fifty-five-minute sessions with late source drills, then sixty-five, then full simulation. The endurance boundary is trained rather than repeatedly observed.

129. Practice Lab: short drills never expose the weakness

Iona looks excellent on twenty-minute tasks. Parents therefore assume stamina is solved. Full-section evidence shows late scope decay. Training must include enough duration to reveal the real environment. Short accuracy and long preservation are different competencies.

130. Practice Lab: early efficiency fixes late stamina

Leonie’s final-third decline disappears after early answer inflation and over-annotation are reduced. She did not need longer training. She needed lower cumulative cost. This case prevents stamina from becoming the default explanation for every late error.

131. Practice Lab: lower cost is not enough

Maren becomes more efficient early and reaches the final third fresher, yet late source attribution still declines. A true endurance component remains. The programme preserves the efficiency gain and adds late source-position training rather than pushing efficiency further.

132. Practice Lab: a duration ladder works

Iona is stable for forty minutes and drifts at fifty. Training holds forty-five until source and scope remain intact, then extends to fifty-five. Several weeks later, the original fifty-minute boundary no longer predicts failure. The learner has evidence that endurance can be developed incrementally.

133. Practice Lab: a duration ladder advances too fast

Leonie moves from forty minutes to full ninety-minute simulations immediately. Late errors multiply and confidence drops. The programme backs down to the last stable duration. Progress resumes once extension is tied to preserved quality rather than determination.

134. Practice Lab: load, not duration, is the limiter

Maren can read ordinary texts accurately for eighty minutes but declines quickly when the final task contains three sources and a graph. The session length is not the main issue. Source-switch and multimodal load under duration is. Training keeps the length constant and increases source complexity gradually.

135. Practice Lab: duration, not load, is the limiter

Iona handles dense source work well at minute twenty but makes the same errors on simple text after seventy. The endurance boundary is broad. Graduated duration and low-load late controls become more important than harder materials.

136. Practice Lab: passage resets remove carryover

Leonie’s late narratives inherit the anxious reading mode of a difficult preceding argument. A five-second passage reset—old queue parked, new text, new voice, new reading resolution—reduces rereads. The improvement comes from state transition rather than more comprehension instruction.

137. Practice Lab: resets become a distraction

Maren turns the reset into a long ritual with notes, breathing counts and repeated clock checks. The routine consumes time and attention. It is compressed to three actions: park, time state, new source. Endurance support must remain low-load.

138. Practice Lab: confidence resets protect simple questions

Iona misses an ambiguous inference and then treats later direct questions as equally uncertain. Training requires each question to rebuild confidence from its own evidence. Simple later items stop inheriting the emotional difficulty of the earlier one.

139. Practice Lab: source labels are enough

Leonie initially receives a complex multi-source worksheet with colours and elaborate notes. Testing shows a simple A/B/C label plus shared axis is enough to preserve late ownership. Extra support is removed. The lower-load system survives longer.

140. Practice Lab: source labels are not enough

Maren still blends two sources despite labels because their positions are similar. The repair adds one distinct identity phrase per source: “A cost cautious”, “B evidence cautious”. Endurance support becomes slightly richer because the simpler version did not carry enough state.

141. Practice Lab: broad reading reduces late effort

Iona spends several weeks reading longer essays outside test conditions. Unfamiliar syntax and discourse moves become less expensive. Later timed sessions show a smaller early–late gap even before specific endurance drills increase. Language familiarity has lowered the cumulative processing cost.

142. Practice Lab: broad reading does not solve one late mechanical failure

Leonie’s general reading improves, but she still forgets digital flags late. The remaining problem is not language stamina. It is interface workflow. Training becomes more specific rather than simply assigning more reading.

143. Practice Lab: subjective fatigue rises while accuracy stays stable

Maren reports feeling exhausted in the final third, yet process and score remain stable. The data do not support a performance problem. Training does not chase the feeling with more endurance work. She learns that effort sensation can rise without loss of competence.

144. Practice Lab: subjective fatigue is absent while errors rise

Iona says she feels fine, but late source and scope errors increase consistently. The script overrides the self-report. Endurance training targets the observable decline. Feeling fresh is not a sufficient performance measure.

145. Practice Lab: the learner improves by reducing internal commentary

Leonie silently narrates every decision in full sentences. This is useful in lessons but expensive in a long paper. She compresses self-talk to short triggers. Late accuracy improves because less working memory is consumed by explaining a process she already knows.

146. Practice Lab: a risk card becomes permanent support

Maren relies on a late-paper card saying source, scope, complete. After several weeks, the routine remains stable without looking at the card. The support is faded. Endurance training should reduce dependence as automaticity grows.

147. Practice Lab: a risk card is faded too early

Iona removes the support after one successful session. Scope drift returns late. The card is restored for several delayed transfer sessions, then faded gradually. Immediate success is not enough to prove endurance automation.

148. Practice Lab: the final third becomes ordinary

Leonie once needed special reminders after minute fifty. After graduated training, her final passage uses the same compact map, local evidence search and answer contracts as the first. She still feels the accumulated effort, but the process no longer changes. This is the target state.

149. Practice Lab: the final third remains weak only on one genre

Maren’s general stamina improves, but late argumentative passages still produce scope errors. The programme narrows. General duration work moves to maintenance; late argument architecture becomes active. Endurance training evolves as the failure signature becomes more specific.

150. Practice Lab: the exam is long, but the learner’s system is longer

Iona completes a full representative section after several months of graduated endurance work. One question is difficult, one source switch is awkward and the final passage is dense. Yet her source labels remain accurate, scope does not broaden, multi-part answers stay complete and checking remains selective. The paper is still demanding. The difference is that the operating system outlasts the demand.

Technical Appendix A: Early–middle–late endurance dashboard

A useful endurance dashboard compares the same few variables across the first, middle and final thirds of a long practice session. The goal is not to collect every possible metric. It is to identify whether the learner’s operating system changes with duration. Good candidates are evidence-region accuracy, source ownership, quantifier fidelity, answer completeness, broad-reread count, response length and confidence calibration.

For each variable, record the early state, the late state and whether the difference is meaningful enough to change training. Example: source attribution 95% early, 70% late across three comparable paired-source tasks; answer length unchanged; rereads stable. That pattern points strongly toward source endurance rather than general fatigue. Another learner may show stable source ownership but late answer length doubling, suggesting uncertainty cost.

Context matters. Mark whether the late task was genuinely more difficult, whether the learner was carrying time debt, whether a confidence shock occurred earlier and whether the interface changed. The dashboard should help separate pure duration effects from task difficulty and disruption. Endurance evidence is strongest when late deterioration appears across fresh work under reasonably comparable conditions.

The dashboard should also record positive preservation. If the learner feels more tired but source, scope and completeness remain stable, that is a success. Endurance training should not aim to erase the sensation of effort. It aims to preserve performance despite accumulating effort.

Technical Appendix B: Stamina failure taxonomy

• Representation fatigue: passage maps become weaker, chronology blurs, source identities merge.
• Retrieval fatigue: local search becomes broad rereading, evidence corridors widen, wrong regions are selected.
• Reasoning fatigue: inference bridges enlarge, comparison axes weaken, evaluation becomes generic.
• Scope fatigue: quantifiers broaden, conditions disappear, certainty rises beyond evidence.
• Expression fatigue: answers become incomplete, overly long, repetitive or semantically distorted.
• Decision fatigue: thresholds drop, guessing rises, objective stop rules disappear.
• Recovery fatigue: one hard item continues affecting later work after it has been skipped.
• Switching fatigue: discourse frames or source states carry incorrectly into the next task.
• Mechanical fatigue: scrolling, transfer, numbering, flags or submission become less reliable.
• Interface-specific fatigue: the decline appears only in digital or only in paper conditions.

Several categories can coexist, but the training target should usually begin with the earliest repeated mechanism. If the passage map collapses first and broad rereading follows, representation may be the upstream endurance problem. If the map remains sound but certainty language becomes extreme, scope is more direct. The taxonomy prevents every late error from being labelled “poor concentration”.

Technical Appendix C: Endurance decision tree

1. Does the skill work while fresh? If no, repair the component first; this is not primarily stamina.
2. Does it fail only when the learner is behind schedule? If yes, route first to timing and early efficiency.
3. Does it fail after one difficult question but not after smooth work? Recovery and confidence shock may be primary.
4. Does it fail at roughly the same late position across comparable sessions? A genuine duration boundary is more plausible.
5. Does it fail only in one genre, modality or source configuration? Train that conditional endurance rather than general duration.
6. Does the learner become slower or faster late? Select the intervention according to uncertainty inflation or threshold collapse.
7. Which control fails first? Map, source, evidence, scope, completeness, checking or mechanics.
8. Can that control be preserved with a small external support? If yes, use it temporarily and fade.
9. Does the repair survive added duration, added load and delay? If yes, recombine into full simulation.

The decision tree distinguishes skill absence from endurance failure. A learner who cannot compare two sources at minute ten does not need late-source training yet. A learner who compares accurately at minute ten and fails only at minute fifty does. This protects training from using duration as a substitute for component instruction.

Technical Appendix D: Duration ladder

Begin with the longest duration at which the target process remains stable. Suppose scope is accurate through forty minutes and drifts around fifty. The first training level might be forty-five minutes with late scope-sensitive questions. Once the learner preserves scope across several fresh sessions, move to fifty-five. The ladder should extend the stable zone rather than repeatedly expose the collapse zone.

At each level, protect one or two control variables. For source stamina, preserve source ownership and comparison symmetry. For answer stamina, preserve completeness and normal response length. For evidence-search stamina, preserve local retrieval and reread count. Total score can be recorded, but progression should depend on the mechanism being trained.

If performance deteriorates sharply at the next step, return to the previous stable level and add component fluency or lower-load routines. Do not interpret the setback as lack of effort. The ladder has identified the current endurance boundary. Training should strengthen the process there before extending again.

Eventually, duration should reach or slightly exceed the representative examination demand during selected sessions. Once that level is stable, maintenance can use less frequent full-duration work combined with late-position drills and ordinary mocks. The learner does not need to prove maximum endurance every week.

Technical Appendix E: Load ladder

Duration and difficulty should not always rise together. A load ladder keeps the session length stable and increases one cognitive dimension at a time. Level one may use one clear source. Level two adds denser syntax. Level three adds a second source. Level four adds a chart. Level five adds ambiguous inference or evaluation late. This shows which load dimension actually breaks endurance.

Load can also be created through switching rather than difficulty. Narrative → argument → chart requires three discourse resets. If late errors appear mainly after several switches, train transition routines. Another learner may handle switching but fail when evidence is distributed across distant paragraphs. The ladder should match the observed failure signature.

The load ladder prevents a common mistake: assigning “harder and longer” practice simultaneously, then calling the resulting failure poor stamina. Controlled progression creates useful information about what the learner can currently preserve and what needs automation.

Technical Appendix F: Passage-boundary reset protocol

A passage reset can be compressed to four steps: park unresolved state from the previous passage; confirm current time and queue; clear the previous source identity and reading mode; orient to the new text. This should take seconds. Its job is to prevent state contamination.

The reset should not become a performance ritual that the learner fears missing. It is a lightweight transition habit. Early training may use visible cues. Later, the learner simply experiences a clean start at each passage boundary. The strongest sign of automation is that a difficult previous passage no longer changes how the next ordinary passage is read.

Source-heavy exams benefit especially from the clear step. The learner can finish Source Set 1, preserve any flag, and then explicitly start Source Set 2 with fresh ownership labels. This reduces blending and frees working memory from maintaining irrelevant old voices.

Technical Appendix G: Low-load late-paper operating card

A late-paper card should contain no more than three to five active controls. Examples: map → source → scope → complete; or local evidence → smallest inference → stop. The card is personalised from the learner’s late failure signature. It is not a universal checklist.

Use the card only in training at first. If it reduces late errors, retain it for several varied sessions, then fade it. If the learner starts reading the card constantly, it has become another source of load. The purpose is to externalise the most fragile controls until they become habitual.

Retire controls that become stable. A learner who no longer loses source identity does not need to keep “source” on the card forever. Replace only when a new repeated late mechanism appears. The card should get shorter as stamina improves.

Technical Appendix H: Teacher stamina protocol

1. Establish that the target skill works while fresh.
2. Collect at least several long-session samples and identify the first late drift.
3. Separate duration, time debt, difficulty and confidence-shock effects.
4. Choose one late-vulnerable mechanism rather than training “stamina” generically.
5. Reduce unnecessary early cognitive cost that may be creating the late symptom.
6. Place the vulnerable skill near the current endurance boundary.
7. Use a temporary low-load support if necessary.
8. Extend duration or load gradually while preserving a defined accuracy control.
9. Retest across fresh genres, media and question wording.
10. Integrate recovery and passage-boundary resets.
11. Return to full simulation only after the target process survives late-position work.
12. Move endurance work to maintenance once early–late differences narrow.

Teacher feedback should describe preservation rather than effort. “Your Source B attribution stayed accurate in the final twenty minutes” is more useful than “good concentration”. If the process failed, name the first late change: “Your map stayed sound, but your comparison lost symmetry after minute fifty.” This turns endurance into something the learner can understand and train.

Teachers should also resist extending sessions merely because the examination is long. Full-duration work has high diagnostic value but high opportunity cost. Use it when it can test integration or a new endurance boundary. Use shorter targeted late-position work when the same mechanism can be trained more efficiently.

Technical Appendix I: Self-study stamina protocol

A self-studying learner can compare two or three longer sessions. Record when the first process change appeared, not only when marks fell. Did rereads widen? Did answers lengthen? Did source labels disappear? Did confidence fall? The repeated earliest change is the best starting hypothesis.

Choose one repair. If late answers become incomplete, use an answer-contract cue in the final third. If late evidence search broadens, preserve the passage map and force local search first. If late uncertainty causes overchecking, use objective release criteria. Do not change several endurance behaviours at once.

Retest at the same duration before extending. A learner should be able to say, “At sixty minutes, I used to lose source identity; now I can preserve it on fresh material.” That is stronger evidence than merely surviving a longer session. Only then should duration or load increase.

Self-study should stop before repeated exhaustion makes the data meaningless. A training session that is unusually poor because the learner is ill, sleep-deprived or heavily distracted may not provide a clean endurance test. Record context and avoid overinterpreting one outlier.

Technical Appendix J: Parent support for reading stamina

Parents can help by asking what changes late rather than simply telling the learner to “build concentration”. Useful questions include: “What do you still do well at the end?” “What starts going wrong first?” “Does it happen because you are behind on time?” “Does one hard question affect later work?” “What late skill are you training this week?”

Avoid making every study session longer. More duration is not automatically more stamina. Targeted late-position training, broad reading and efficient early routines may produce better progress. Full-duration sessions should have a reason and should be reviewed afterward.

Parents can also protect recovery between major practice sessions. Endurance training works when the learner can produce representative high-quality effort, not when every evening becomes an exhausted simulation. Sustainable consistency is more useful than occasional marathon volume.

If persistent physical symptoms—such as significant headaches, visual discomfort or other health concerns—appear during reading, treat those as separate from an exam-strategy problem and seek appropriate professional advice rather than attempting to solve them with stamina drills.

Technical Appendix K: Six-week stamina build

Week 1 — Baseline. Run one representative long session and one shorter stable session. Compare early and late process. Identify the first repeated drift and rule out obvious time debt or task-difficulty explanations.

Week 2 — Reduce cumulative waste. Remove early over-annotation, answer inflation, broad rereads or unnecessary checking that may be draining later performance. Keep the target endurance mechanism unchanged for comparison.

Week 3 — Late-position isolation. Place the vulnerable operation near the current boundary. Use a temporary low-load support: source labels, answer contract, scope cue or local-search routine. Train accuracy before extending duration.

Week 4 — Duration extension. Add a modest amount of time while keeping task difficulty ordinary. Preserve the target control. If quality falls, return to the previous stable duration.

Week 5 — Load and switching. Add one realistic complexity dimension: denser source, paired text, multimodal switch or controlled hard-question recovery. Confirm that the learner can reset and restore normal execution.

Week 6 — Full integration. Run a representative mock. Compare early, middle and late process, not only total marks. If the target endurance signature no longer appears, move it to maintenance and identify any remaining conditional weakness.

Technical Appendix L: Twelve-week endurance programme

Weeks 1–2: diagnose the endurance signature and stabilise measurement. Weeks 3–4: reduce unnecessary cumulative cost and automate low-load routines. Weeks 5–6: train vulnerable skills late with controlled duration. Weeks 7–8: extend duration gradually and add source or modality switching. Weeks 9–10: integrate hard-question recovery, passage resets and final-third checking. Weeks 11–12: use representative full simulations and taper explicit supports.

The twelve-week programme should not simply lengthen every week. Some weeks may hold duration constant while the learner masters a new load condition. Others may reduce difficulty while extending time. The objective is preservation of the operating system, not a monotonically rising clock.

Broad reading and language development can run throughout the programme. They reduce the cost of vocabulary, syntax and discourse processing, making endurance easier to sustain. These foundations are particularly important when the learner’s late decline reflects cumulative language-processing cost rather than strategy alone.

Technical Appendix M: Full walkthrough — Maren’s apparent stamina problem begins early

Maren’s final-third accuracy is poor. She loses comparison symmetry, misses one summary component and makes two careless-looking scope errors. The first diagnosis is low stamina. A timeline shows something more specific. During the opening forty minutes she writes long answers, annotates every paragraph and checks direct questions twice. By the midpoint she is already behind and carrying a large amount of visual and cognitive clutter.

Cycle 1 — remove early tax. Annotation is reduced to paragraph jobs and turns. One-mark answers are shortened to the scoring core. Reassurance checking is removed from low-risk questions. Maren reaches the final third on time with fewer pages of notes. Late accuracy improves immediately, but one source-attribution weakness remains.

Cycle 2 — isolate genuine endurance. Paired-source comparison is now tested at minute twenty and minute sixty. Early work is excellent; late work still blends ownership. The source endurance problem is real. A late risk card uses only “A/B + axis”. The task is practised near the boundary on fresh source pairs.

Cycle 3 — extend duration. Once source ownership holds at sixty minutes, the session extends gradually. A controlled hard question is inserted before the final source set. Passage-boundary resets and preserved state prevent confidence shock from contaminating the comparison. Full simulations show a much narrower early–late gap.

Maren’s case shows why stamina should not be diagnosed from the location of the symptom alone. Part of the late collapse was produced by expensive early behaviour. Only after that cost was removed could the genuine late vulnerability be seen clearly.

Technical Appendix N: Full walkthrough — Iona’s strong skills fail only after switching load

Iona can sustain long narrative and argumentative reading separately. Her difficulty appears when a long exam moves from prose to paired sources, then to a graph, then back to prose. She reports feeling mentally tired, but duration alone does not explain the errors. The failure clusters after representation switches.

Cycle 1 — isolate switching. Sessions keep duration moderate but include deliberate mode changes. Iona practises carrying the active question across sources: claim stays active, source job changes, evidence is collected, answer state resumes. Micro-resets separate old and new discourse frames.

Cycle 2 — add duration. The same switches move later in the session. Source labels and mode jobs remain visible. Iona learns that a graph is not a new examination; it is another source answering the current claim. Switching cost falls and late evidence errors decline.

Cycle 3 — integrate recovery. A difficult prose inference is placed before the final multimodal task. Iona parks it with a breadcrumb and enters the graph set cleanly. The final task no longer inherits uncertainty from the earlier block. Her stamina improves because switching and recovery are now low-load routines.

Iona’s case demonstrates that “long exam stamina” can actually be representation-switch endurance. The clock was not the only stressor. The number of mental frame changes mattered. Once those transitions became cheaper, the same duration became manageable.

Technical Appendix O: Handoff to SETC-SEC-COMP-150

The final reserved spoke in this planned sequence is SETC-SEC-COMP-150 — Reading Comprehension Exam-Day Strategy | From the First Page to the Final Answer. 140 owns endurance training before the exam: how to make the system survive duration. 150 will own the integrated day-of-performance route itself: opening the paper, establishing the first passage, managing transitions, protecting the final third, using the reserve and closing the paper.

The boundary is important. 150 should not reteach every stamina drill, timing method or recovery technique. It will orchestrate the already-trained lane under one examination event. 140 supplies the endurance capacity that 150 assumes is available.

151. How SETC-SEC-COMP-140 connects to the lane

This page owns comprehension-specific endurance: identifying late-paper drift, separating fatigue from time debt and recovery shock, reducing cumulative cognitive cost, preserving source/scope/completeness under duration, training vulnerable operations late, using duration and load ladders, and making passage-boundary resets automatic. It receives timing controls from 030, recovery from 070, transfer design from 090, simulation evidence from 100, final checking from 110, speed–accuracy controls from 120, and plateau routing from 130.

Return to SETC-SEC-COMP-000 for the complete comprehension performance system. The next reserved spoke is SETC-SEC-COMP-150, the exam-day orchestration layer.

152. Final principle: stamina is the ability to preserve the rules that make you accurate

Long examinations do not create entirely new reading problems. They expose whether good reading survives cumulative load. The learner must still identify who says what, what the question wants, where the evidence sits, how strong the inference is, how many components the answer needs and when an uncertain item should be released.

Maren protects structure and completeness. Iona protects evidence, source and scope. Leonie protects the continuity of the operating system from passage one through the final minutes. Endurance is mature when these protections remain ordinary even after the learner no longer feels fresh.

READING STAMINA IS NOT PROVING THAT YOU CAN SUFFER THROUGH A LONG PAPER. IT IS TRAINING SO THAT THE PAPER CAN BECOME LONG WITHOUT YOUR STANDARDS BECOMING SMALL.

Advanced Stamina Lab A: same duration, different cognitive history

Maren and Iona both reach minute sixty. Maren has moved through ordinary questions efficiently; Iona has carried three unresolved flags and spent several minutes in repeated option comparison. Their clock position is identical, but their cognitive history is not. Iona’s late decline should not be interpreted as pure duration. Her programme targets unresolved-state load and reassurance loops before asking her to endure longer. Stamina depends on what the learner has carried through the hour, not only the hour itself.

Advanced Stamina Lab B: same fatigue feeling, different performance

Leonie and Maren both report feeling very tired after seventy minutes. Leonie’s source attribution and scope remain stable; Maren’s quantifiers broaden and multi-part answers become incomplete. The subjective report is useful but not diagnostic by itself. Leonie may need confidence that effort sensation is compatible with strong performance. Maren needs late scope and answer-contract training. Endurance decisions are made from the combination of felt effort and observable drift.

Advanced Stamina Lab C: time pressure disappears but late errors remain

Iona receives a generously timed long practice section. She is never behind, yet source ownership and certainty still degrade in the final third. This clears time debt as the primary explanation. The endurance intervention moves to working-memory externalisation and late source/scope work. A learner can have enough minutes and still lose decision quality through accumulated processing load.

Advanced Stamina Lab D: time pressure disappears and late errors disappear too

Leonie’s late decline vanishes when the same paper is completed with a slightly larger time budget. Her component accuracy remains strong. The evidence points back toward pacing and speed–accuracy conversion rather than general stamina. 030 and 120 take priority. 140 avoids claiming a fatigue problem when the learner’s standards are stable until urgency changes behaviour.

Advanced Stamina Lab E: the first drift is answer length, not accuracy

Maren’s marks remain stable for now, but late one-mark answers double in length and checking time rises. This is an early endurance signal. Training introduces late response-contract drills before the longer answers create time debt and later accuracy loss. The best endurance work often begins before marks visibly fall.

Advanced Stamina Lab F: the first drift is evidence-search width

Iona still gets late questions correct, but local search becomes paragraph-wide rereading. Her map is no longer being trusted. The programme moves evidence-corridor prediction into the late phase. Preserving local retrieval now may prevent a future timing and stamina cascade. Search width can be a leading indicator even while answers remain correct.

Advanced Stamina Lab G: the first drift is confidence calibration

Leonie’s late answers remain mostly correct, but confidence ratings become uniformly low. She checks them repeatedly and begins changing correct responses. Her endurance limit sits in calibration rather than comprehension. Late practice requires objective evidence-based release: if command, support, scope and completeness are satisfied, the answer can be released despite lower subjective certainty.

Advanced Stamina Lab H: the first drift is source-map compression

Maren’s early source maps contain distinct positions. Late maps become only “A / B”, and similar views blur. Training gives each late source one identity phrase tied to the comparison axis. The support is still compact but carries more discriminating information. Endurance externalisation should preserve the information actually at risk, not use the smallest possible note for its own sake.

Advanced Stamina Lab I: late accuracy fails only after multiple switches

Iona can work for an hour on one long text family without difficulty. Her errors appear after prose → chart → paired prose → summary transitions. The stamina programme counts frame changes as load. Micro-resets and mode-job labels are trained at moderate duration before being moved later. Endurance can be limited by switching density rather than continuous reading time.

Advanced Stamina Lab J: late accuracy fails only after unresolved queue growth

Leonie’s ordinary late accuracy is strong when only one item is flagged. When four unresolved questions accumulate, later source and scope errors rise. The queue itself is carrying cognitive load. Training improves minimum engagement and skip thresholds so only genuine blocks enter the queue, and breadcrumbs reduce restart cost. Stamina improves because fewer open loops travel through the paper.

Advanced Stamina Lab K: a long easy passage is less tiring than a short dense one

Maren assumes endurance should correlate with word count. Yet a short argument with multiple concessions and sources causes more late drift than a longer narrative. The programme begins tracking conceptual density and relation count, not just length. Endurance training becomes better calibrated because cognitive load is measured by processing demand rather than page size alone.

Advanced Stamina Lab L: the final passage is easier but performance is worse

Iona answers an ordinary late passage worse than a denser early one. Because difficulty cannot explain the decline, endurance becomes more plausible. Early–late matched items confirm the pattern. Training holds difficulty steady and extends duration until the ordinary late text is treated with the same standards as the ordinary early text.

Advanced Stamina Lab M: a difficult final passage is mistaken for fatigue

Leonie’s final-third score is lower, but the late passage is consistently the most complex in the practice set. When similar difficulty is placed early, the same errors appear. The primary problem is component difficulty, not stamina. The programme trains the hard question family while fresh before returning it to the late position.

Advanced Stamina Lab N: late scope improves with one-word checking

Maren’s late answers are conceptually good but overstate quantifiers and certainty. A full re-read would be expensive. Her low-load endurance repair is a one-word risk scan: all, only, proves, always, never. She checks these high-risk scope markers before releasing selected late answers. Small verification protects proposition fidelity without creating another fatigue burden.

Advanced Stamina Lab O: late completeness improves with answer-slot counting

Iona’s final questions lose one component. She begins using a two-second pre-writing slot count: two reasons = two slots; compare = A + B; device effect = device + effect. The routine is faster than discovering incompleteness during final checking. Endurance improves by moving a tiny completeness control upstream.

Advanced Stamina Lab P: late inference improves when alternatives are capped

Leonie generates many possible readings late because uncertainty feels dangerous. Training allows one primary interpretation and at most one serious competitor. If the primary interpretation has stronger evidence and lower assumption cost, she closes. Limiting option generation protects reasoning quality and cognitive resources simultaneously.

Advanced Stamina Lab Q: late vocabulary improves when unknown words become placeholders

Maren encounters a technical noun in the last passage. Instead of resolving it completely, she labels it Entity X and tracks what X causes and what changes it. The relationship is enough for the current questions. Placeholder tolerance keeps unfamiliar vocabulary from consuming the final cognitive budget.

Advanced Stamina Lab R: late syntax improves when one sentence stays local

Iona hits a complex conditional sentence and begins to feel the whole passage is difficult. Her trained repair isolates the sentence, identifies the main proposition and condition, then returns to normal pace. The local problem does not become a global state. Endurance includes limiting the geographic spread of difficulty.

Advanced Stamina Lab S: the learner carries a previous narrator into a new text

Leonie moves from one narrative passage to another and accidentally attributes the first narrator’s sceptical tone to the second. The passage reset now includes “new voice”. This seems obvious while fresh, but late source and narrator states can persist implicitly. Clearing ownership at boundaries prevents cross-text contamination.

Advanced Stamina Lab T: the learner carries argumentative caution into a direct graph task

Maren finishes a nuanced argumentative passage and then hedges a graph answer that directly shows a decline. The previous task’s uncertainty level has carried over. A mode-job reset reminds her that each question rebuilds certainty from its own evidence. The graph can “show” what the argument could only “suggest”.

Advanced Stamina Lab U: the learner carries urgency into a new passage after one overrun

Iona overruns one question and rushes the entire next passage despite the remaining schedule still being recoverable. A passage-boundary time-state check recalibrates urgency. Instead of “I am behind”, she uses the actual debt and adapts proportionally. Endurance improves because urgency no longer becomes a persistent cognitive mode.

Advanced Stamina Lab V: the learner carries excessive caution after a wrong answer

Leonie discovers that an early answer was wrong and responds by checking every later item repeatedly. Her paper becomes slower and more tiring. Recovery training reframes the error as local information: correct the dependency chain, then restore normal standards. One wrong answer should not globally change the learner’s trust threshold.

Advanced Stamina Lab W: the final summary is protected by earlier efficiency

Maren’s summary used to collapse late because earlier answers were long and heavily checked. After answer inflation falls, she reaches summary with more time and mental capacity. Selection and compression stabilise. The endurance intervention occurred before the summary itself. Protecting critical late tasks can require reducing cumulative earlier cost.

Advanced Stamina Lab X: the final summary still fails despite good time state

Iona reaches summary with adequate reserve but selects duplicates and drops conditions. The timing explanation is cleared. Late summary-specific endurance remains. Training uses summary selection and paraphrase drills after longer reading blocks until proposition architecture survives load.

Advanced Stamina Lab Y: a digital interface hides late flags

Leonie preserves cognitive accuracy but fails to reopen one late digital flag because it sits outside the current screen. Her endgame adds a global flagged-items view before submission. Endurance is partly the ability to use the interface as external memory when internal memory is least reliable.

Advanced Stamina Lab Z: a paper script makes late state visible

Maren’s paper annotations allow her to see the location of a skipped item, the source labels and a summary point set physically. Her endurance is better on paper than digital. The digital programme recreates those external-memory functions with flags, landmarks and concise notes rather than assuming the cognitive skill should transfer automatically across media.

Advanced Stamina Lab AA: fatigue resistance improves when the skill becomes automatic

Iona once had to consciously ask whether a comparison had both sources. After repeated structured practice, the A + axis + B architecture becomes automatic. Late in long sessions, the comparison remains complete without extra prompting. The endurance gain did not come from greater willpower; it came from lowering the conscious cost of correct execution.

Advanced Stamina Lab AB: automaticity becomes dangerous when the rule is wrong

Leonie automatically labels any critical tone as hostile. Because the rule is fluent, fatigue makes it even more dominant. The concept must be repaired before endurance training. Stamina preserves trained processes; it cannot distinguish good automation from bad automation. Component accuracy must precede fatigue resistance.

Advanced Stamina Lab AC: endurance gains appear first as fewer cascades

Maren’s late question accuracy is not yet higher overall, but one difficult passage no longer causes the next passage to collapse. The endurance system has improved restoration. This is meaningful progress even before every late component strengthens. Fewer cascades raise the performance floor and make remaining fatigue mechanisms easier to diagnose.

Advanced Stamina Lab AD: endurance gains appear as stable response length

Iona’s total score is similar, but late one-mark answers no longer expand into paragraphs. She is preserving answer contracts under uncertainty. Time debt shrinks and final checking becomes more targeted. Stable response length is a useful leading indicator when answer inflation was the earlier endurance signature.

Advanced Stamina Lab AE: endurance gains appear as stable source attribution

Leonie still feels tired in the final source set, but A/B/C ownership remains correct across several simulations. The old source-blending failure has disappeared. Source endurance can move to maintenance even if another late weakness, such as checking or summary, remains active.

Advanced Stamina Lab AF: endurance gains appear as a narrower early–late gap

Maren’s early accuracy stays high while late accuracy rises gradually over six weeks. The goal is not necessarily identical performance in every segment because task difficulty varies. The meaningful pattern is that comparable late tasks increasingly resemble early execution in source, scope and completeness. The endurance gap is closing.

Advanced Stamina Lab AG: longer practice makes performance worse because recovery is insufficient

Iona begins doing full-duration papers almost every day. Her late accuracy deteriorates and review quality becomes superficial. The programme reduces full simulations and restores targeted component work. Endurance is not built by continuous depletion. Training requires enough recovery that long sessions remain representative and analytically useful.

Advanced Stamina Lab AH: shorter high-quality sessions build a longer system

Leonie alternates focused late-position drills with occasional full simulations. Source ownership, scope and switching become more automatic. When she returns to a long mock, the final third improves despite doing fewer marathon sessions. Endurance was built through component efficiency and targeted preservation, then verified through duration.

Advanced Stamina Lab AI: one week of improvement is not enough to fade support

Maren’s late risk card works immediately. She wants to remove it after two sessions. Instead, the support remains through different genres, a delayed retest and one full simulation. Only after the routine survives surface change is the card faded. Endurance automation requires persistence across time and context, not one good day.

Advanced Stamina Lab AJ: support can be faded by information, not by time

Iona’s late source note begins as “Source A cautious because cost; Source B cautious because evidence; axis = reason for caution”. As stability improves, it shrinks to “A cost / B evidence / caution”. Later only A/B labels remain. The support fades by compressing information while preserving function, rather than disappearing abruptly.

Advanced Stamina Lab AK: final-third checking becomes more efficient than early checking

Leonie once believed tiredness required checking more. Training teaches the opposite: late checking must be more selective. She scans blanks, partials, known scope risks and high-value flags instead of rereading secure work. The final minutes become less cognitively expensive even though the learner is less fresh.

Advanced Stamina Lab AL: a final-third risk list is too long

Maren carries eight reminders into the last passage. Monitoring the list competes with reading. The risk list is reduced to source, scope and complete—the three recurrent late failures. Other controls are already sufficiently automatic. Endurance support should be prioritised, not comprehensive.

Advanced Stamina Lab AM: a final-third risk list is too vague

Iona’s card says “focus” and “be careful”. These cues do not specify behaviour. They are replaced with “local evidence”, “condition”, “both sources”. The new cues point to observable operations. Stamina improves when the learner knows what to preserve, not merely that preservation matters.

Advanced Stamina Lab AN: the learner becomes dependent on a clock check

Leonie checks the clock every few minutes because she fears late collapse. The monitoring itself fragments attention. Timing checkpoints are returned to 030’s planned boundaries, and stamina work focuses on process preservation between them. Endurance should not require constant vigilance about endurance.

Advanced Stamina Lab AO: the learner avoids difficult passages to protect stamina

Maren starts selecting easy practice because longer hard texts feel draining. Her endurance appears better, but the environment no longer represents the target. Training returns to controlled challenge: ordinary difficulty at longer duration first, then harder late tasks. Stamina must transfer to the examination’s real demand, not merely to a comfortable version of it.

Advanced Stamina Lab AP: the learner uses full papers when one late drill would do

Iona’s only recurring endurance weakness is late comparison symmetry. She keeps sitting full sections to practise it. The programme instead places ten-minute comparison tasks after forty-five minutes of ordinary work. The target receives more focused repetitions with less total fatigue. Full mocks return for confirmation, not every exposure.

Advanced Stamina Lab AQ: the learner uses isolated drills after integration is ready

Leonie’s late scope control is stable across mini-sets, genres and delays. She continues isolated drills because they feel safe. The next step is full integration: scope must survive source switching, ambiguity, time state and final checking. Endurance training eventually has to re-enter the complete paper.

Advanced Stamina Lab AR: the learner’s final-third weakness moves

Maren repairs late source attribution. Weeks later, the main endurance loss becomes answer completeness. The programme changes. Stamina is not one permanent capacity that is either present or absent. Different controls can become limiting as others automate. Endurance diagnosis should be repeated as the learner improves.

Advanced Stamina Lab AS: the learner no longer needs a special final-third strategy

Iona once used several late-specific supports. After sustained training, the same compact passage map, source control, answer contracts and recovery rules work from beginning to end. Her final-third card is retired. This is the strongest endurance outcome: not a special survival mode, but ordinary comprehension that remains ordinary for the whole examination.