CSIR NET Part A Syllabus 2026: 360 Official Questions Analysed

The actual Part A syllabus: official scope versus a useful study map

Part A is common to the science subjects, so the official scope is written as abilities rather than as a textbook index. “General Aptitude” does not mean a second science-subject paper. You may meet a physics-flavoured graph, a biological population table or a simple geometry diagram, but the test is whether you can reason with the information supplied. Our 13 types below split that broad scope into trainable methods. The split is editorial; it is not an NTA chapter list.

For a beginner, the map suggests an order: get comfortable with number properties, ordinary arithmetic and geometry, then practise reading graphs and formal deduction. But “largest in this sample” should not become “ignore the rest.” The 18 sets differ, and the smaller types still cost marks when they appear. A candidate strong in geometry but weak in Venn problems should repair Venn reasoning first despite the aggregate count.

The label also matters when you review a mistake. If you missed a chart question because you divided by the wrong baseline, tag it as a data-interpretation percentage error, not as a general “graph weak” result. If your answer to a family-relation question relied on an unstated assumption, tag deduction. A practical topic plan emerges from repeated error causes, not a list of formulas you have already memorised.

How we counted the 360 questions—and what the chart cannot say

We began with 25 subject-paper PDFs in the CSIR-HRDG Question–Answer Keys Archive: December 2023, June 2024, the December 2024 cycle that sat in February–March 2025, and June 2025. Part A in subject papers held in the same shift often repeats exactly. Counting every copy would make a shift with more published subject PDFs look more important, even though the questions were identical. We therefore compared Part A sets and retained each identical same-shift set once. This leaves five sets in December 2023, six in June 2024, five in the December 2024 cycle and two in June 2025: 18 sets of 20 questions.

A few choices are judgement calls. A question about a graph of moving objects could be coded as graphical interpretation or everyday science reasoning; a geometry question that turns on a divisibility trick could fit either geometry or number properties. We use the primary method a learner needs to get to the answer, one code per item. This counts question stems appearing in the official papers, not scored answers: a later key change or dropped item does not turn a printed question into a prediction. Repeated questions in different shifts or cycles are counted again because they are actual separate sittings; exact copies within the same shift are not. These rules make the chart reproducible, but no frequency chart can tell you what the next paper will contain.

The 18 distinct Part A sets retained in the audit

Cycle-by-cycle counts, so the bars have a table fallback

The June 2025 column must not be read as a collapse in a topic: it contains two distinct sets, while June 2024 contains six. If you want a fair per-set comparison, divide a column by the number of retained sets shown in the inventory. Even then the sample is too short to extrapolate a trend. The strongest use of the dataset is to spot methods you should be able to perform under time pressure, not to manufacture a “guaranteed” question list.

Turn the syllabus into a method: six original worked mini-examples

The short examples that follow are newly written practice illustrations, not copied CSIR paper questions or claims about an answer key. Each demonstrates the move behind a family of official-question types. Work each without a calculator, then use an official paper to see how the same method is disguised in a real stem.

1. Number properties: reduce before you expand

Suppose a question asks for the remainder when 99 to the power 12 is divided by 100. Because 99 is −1 modulo 100, an even power leaves 1. Expanding the huge number is wasted work. The transferable move is to rewrite a base near a convenient modulus and inspect whether the exponent is even or odd. In our audit, number properties and algebra were 54 of 360 items, the largest single label, but they covered several different operations—remainders, digits and equations—not one trick.

2. Arithmetic: keep the denominator honest

A sample solution is 20% smaller and then 20% larger. Start with 100; it becomes 80, then 96. The net change is −4%, not zero. The second percentage uses 80 as its base. This is the same reasoning you need for successive discounts, population change or mixture replacement. Write the starting quantity in the margin before doing mental percentages. Arithmetic word problems made 51 of our 360 labels.

3. Geometry: draw the constraint before reaching for a formula

A rectangle has sides 6 and 8. Its diagonal is 10 by the right-triangle relation. If the sides both double, the diagonal doubles to 20; the area instead becomes four times as large. The first question to ask in a geometry stem is whether the figure is scaling a length, an area or a volume. That unit decision often removes a wrong option before calculation. We coded 50 geometry and mensuration items in the sample.

4. Graphs: distinguish absolute change from percentage change

One series rises from 20 to 30, and another from 100 to 115. The second gained 15 units and the first only 10, but the first grew 50% against 15%. A graph question asking for “largest increase” is different from one asking for “largest relative increase.” Read the axis, unit and baseline before comparing. Graph and table interpretation accounted for 33 of our coded questions.

5. Deduction: separate “must” from “may”

If all researchers are graduates and some graduates are chemists, can we conclude that some researchers are chemists? No. Those chemists may be outside the researcher group. Sketch two nested or overlapping circles and try to produce a valid counterexample. Formal deduction, Venn counts and relation logic together formed 31 of our labels, while arrangement and ranking puzzles were another 25.

6. Counting: define the sample space once

A fair coin is tossed twice. The four equally likely ordered outcomes are HH, HT, TH and TT. The chance of at least one head is three of four, since only TT fails. The method is to count the complement when “at least one” makes direct counting messy. In past papers we classified 29 items as counting or probability. The same counting habit helps with digit arrangements and simple selection puzzles.

After each example, look for the specific trigger in the official paper: “remainder,” “relative,” “must follow,” “at least one,” or a unit of area versus length. A revision note should say which trigger made you choose a method and what error would have changed the answer. That is more useful than a page of isolated formulas.

Part A marking, question selection and the first-15 rule

The June 2026 bulletin prints the same Part A offer, attempt limit and marks for Chemical, Earth, Life, Mathematical and Physical Sciences: 20 offered, at most 15 counted, 2 marks right, 0.5 deducted wrong. The subject differences sit chiefly in Parts B and C. If you answer too many questions, NTA says only the first 15 Part A questions answered are evaluated. That means the order of your clicks matters; do not use the 20 questions as a 20-attempt paper and expect NTA to choose your best 15.

There is no separate Part A qualifying cutoff in the June 2026 bulletin. Its merit rule uses cumulative marks from Parts A, B and C. Treat the Part A calculator below as section practice arithmetic, not a pass/fail prediction. The same bulletin prohibits carrying a calculator into the test, so solve the worked examples and official sets by hand.

A blind guess among four equally likely options has an arithmetic expected value of +0.125 marks under +2/−0.5. That observation alone does not mean “guess everything.” Reading a question, excluding choices and entering an answer takes time that could have helped in Part B or C. Worse, an early guess can occupy one of the first 15 Part A slots before you reach a question you know. If you can eliminate one or two choices, the expected value improves under the equal-likelihood assumption, but a personal mock log is a better guide than an abstract average.

A practical two-pass approach is to scan for questions you can solve reliably and quickly, then return to the remaining methods if you still have Part A slots and time. This is a practice framework, not an official recommended order. In your mock review, write down how many correct answers came from fast recognised methods, how many came after long calculation, and whether a guess displaced a later known question. Those three numbers tell you more than a generic “attempt 15 at all costs” slogan.

A diagnostic revision plan that changes with your mistakes

First take one timed official Part A set. For each miss, record the source cycle, subject/shift, question number, our type, why your chosen method failed and the method you would use next time. Do not score a topic as “weak” from one awkward question. Look for repeated causes over two or three sets. Give a skill a higher weakness level when you are repeatedly wrong or consistently too slow, then allocate a modest weekly Part A block that still leaves time for the subject paper.

The default eight-hour example above is deliberately mechanical: it divides hours by observed question counts when every family is set to the same weakness level. It is not a claim that every candidate should spend eight hours on Part A or that the next paper will copy the sample. In the planner, changing “Graphs / DI” from “needs practice” to “often wrong” increases its share; changing a strong family to “confident” reduces its share. Review the allocations against your exam date, subject workload and actual paper errors. If your science subject is the bottleneck, reduce total Part A hours rather than believing any Part A chart can compensate for neglected Part B/C work.

A useful weekly sequence is method repair, then one mixed timed set, then an error review. Method repair means doing a few carefully chosen examples with the solution hidden; the timed set tests whether you can recognise the method amid unfamiliar wording; the review turns misses into the following week’s tasks. If a type stays weak because the underlying mathematics is missing, step back to that prerequisite. If it stays weak only under time pressure, keep the concept and change the decision to skip or return.

Where to get the real papers, and how to use them without copying answers

The paper ledger matters because a past-paper article can accidentally multiply identical sets. If you study a Chemical Sciences and Physical Sciences PDF from the same shift and find the same Part A, treat it as one practice set. You can still practise both subject papers for Parts B/C, but for Part A frequency analysis it is one observation. Our 18-set count already applies that rule.

The June 2026 papers with recorded responses were made available through NTA’s answer-key challenge process, but the public CSIR-HRDG archive set used in this analysis ends at June 2025. We therefore do not mix an unarchived 2026 key or candidate response into the 360-question dataset. When an accessible official question paper and final key are published together, add that cycle as a dated extension and show how the chart changes; do not quietly relabel the old sample “2026 trend.”

What changes for the December 2026 sitting

The December 2026 information bulletin had not been published at this check. Our pattern table therefore reports the latest available June 2026 bulletin, not an invented December rule. CSIR has separately announced a Life Sciences and Biotechnology subject change for December 2026; that concerns the subject syllabus and potentially Parts B/C. Do not assume a new Life Sciences & Biotechnology Part B/C pattern from the June Life Sciences table. This page is about common Part A, and its official rules must be refreshed as soon as NTA issues the December bulletin.

If the new bulletin changes a Part A value, the top answer box, scoring tool, static examples and kit copy must all be checked together. If it does not change the rules, the old question dataset still needs a clear date because the public archive can grow. The “checked” banner at the top tells you which documents this version relied on.

Is there an official chapter-by-chapter CSIR NET Part A syllabus PDF?

CSIR and NTA describe Part A as general aptitude with logical reasoning, graphical analysis, analytical and numerical ability, quantitative comparison, series and puzzles. They do not publish a fixed chapter-by-chapter weightage table. Use the official bulletin and model paper for scope; the 13 types here are ProSyllabus study labels.

How many Part A questions can I attempt?

The June 2026 NTA bulletin offers 20 Part A questions and evaluates a maximum of 15. It says that if more than 15 are answered, only the first 15 Part A answers are evaluated. Check the new December bulletin when released.

Does CSIR NET Part A have negative marking?

Yes, the June 2026 bulletin gives +2 for a correct Part A answer and deducts 0.5 for an incorrect one. Parts B and C differ by subject, so do not copy a Part A rule into those sections.

Is there a separate cutoff for Part A?

No. The June 2026 NTA bulletin says there is no separate cutoff for Parts A, B and C; the merit list uses their cumulative marks. Its category benchmarks concern the combined result, not a required Part A score.

Are the 360-question topic counts official weightage?

No. The papers are official, but the 13-type classification and deduplication are ProSyllabus analysis of 18 distinct Part A sets from December 2023 through June 2025. They describe a sample, not the next paper.

Where are the previous Part A papers?

The CSIR-HRDG Question–Answer Keys Archive links official cycle indexes, which in turn link each subject paper. Part A sits at the start of the relevant paper. Same-shift subject PDFs may contain identical Part A questions.

Is the seven-page revision kit included with a quiz membership?

No membership entitlement is promised here. The CSIR revision kit is a separate digital product; its offer page shows the current price and a free preview of page 1.