Free Study Resources and Reading Plans: 150 Books and Courses That Cost Nothing
The open libraries, open textbooks and open courseware worth building a term on — and how to use each one
id="why-the-free-shelf-is-better-than-it-looks">Why the free shelf is better than it looks
The first thing to unlearn is that "free" and "available" are the same thing. Libraries are full of records: metadata about books you cannot download, catalogue entries for titles whose only copy sits in a reading room in Boston, and links that resolve to a lending queue rather than a file. Search engines flattened all of those into things that looked identical to a public-domain text you could download in ten seconds.
The open shelf is different, and better, for one structural reason: the whole work is often the work. On the open shelf, the thing you get is the thing. A peer-reviewed textbook chapter with the exercises attached. A lecture with the problem sets attached. A paper with the data attached. There is no paywall between you and the end of the document, no "request a copy" email, no library proxy that only works on campus on weekdays.
That has knock-on effects. You can download the PDF and read it on a bus. You can search inside it. You can print the one page you need. You can cite it without a permission slip. You can hand it to a student. For a self-directed learner this is the difference between a plan that survives month three and one that does not.
What has actually changed
Publishing changed. A publisher can now produce a book for zero marginal cost, distribute it to anyone, and still pay the author. That is not charity; it is a better margin than paper.
Institutions changed. A university running a course has already written the readings, set the problem sets and chosen the textbook. Publishing that costs the lecturer nothing and gives the university reach. Open courseware and open textbook collections are institutional by-products that happen to be free.
Public money paid for it. Grants and foundation funding paid for thousands of books to be written and released under open licences. The grant ended; the books did not.
The three things worth knowing before you start
Licences vary, and the differences are practical. A Creative Commons licence controls reuse, not price. CC BY lets you copy, adapt and redistribute with attribution. CC BY-SA asks you to keep the licence on adaptations. CC BY-NC forbids commercial use. CC BY-ND forbids adaptations, so you can share the PDF but not edit it. These matter when you want to print a chapter, translate it, put it in your own course pack, or republish it.
"Free" and "no ads, no tracking, no account" are different promises. Some open materials are hosted on a commercial video platform with trackers attached. That is often the only way they exist, and it is a reasonable trade — but know that you are making it.
Formats differ. A PDF is searchable but not responsive on a phone. An EPUB reflows. HTML is best for screen readers and for copying code out. MP3 or M4B is for listening. A book that exists only as a PDF of a scanned 1962 typescript is legally free and practically useless; the same book in clean EPUB is a different object. Check the format before you commit.
What this guide does with the sources
Each source below is described by what it actually holds, how large it is, what format it uses, what it is good for, and where it stops. Where a source is thin, that is said. Where a title is free to read but not free to download, that is said too, because the difference is invisible from a search results page.
The goal is not a link dump — you already have one. The goal is a plan you can execute: a shelf, a method for checking what is really there, and a schedule.
id="the-twelve-open-libraries-and-what-each-one-actually-holds">The twelve open libraries, and what each one actually holds
Here is the working shelf. Sizes are approximate and change over time, so treat them as orders of magnitude rather than exact counts.
| # | Source | What it actually holds | Size | Formats | Best for |
|---|---|---|---|---|---|
| 1 | Project Gutenberg | Digitised public-domain works, mostly pre-1930 literature, plus later works released voluntarily | ~75,000 ebooks | EPUB, Kindle, plain text, HTML | Reading literature; offline; plain-text analysis |
| 2 | Internet Archive / Open Library | A lending library plus web archive: books (borrowable), films, audio, software, entire websites | ~250 million items | PDF, EPUB, borrow-streamed, MP3, video | Borrowable books; archival research; images and film |
| 3 | OpenStax | Peer-reviewed open textbooks, written by leading academics and peer-reviewed by their peers | ~1,000+ titles (books + lecture notes) | PDF, online, Kindle, print-on-demand | First-year university and high-school courses |
| 4 | Open Textbook Library | A catalogue of open textbooks from many publishers, with reviews and adoption info | ~4,000+ titles | Mostly PDF and online | Comparing textbook options before you commit |
| 5 | BCcampus Open Textbook Collection | Peer-reviewed open textbooks in BC and beyond, in subject groups, with attribution files | ~500+ titles | PDF, DOCX, ebook | Humanities and social-science-heavy courses |
| 6 | MIT OpenCourseWare | Full course materials: lecture notes, problem sets, exams, readings lists, in many cases videos | ~2,000+ courses | PDF, HTML, video | Studying a course you are not enrolled in |
| 7 | OpenLearn (The Open University) | Free course materials built for distance learners, with forums and activity prompts | ~300+ courses | PDF, HTML, video, quizzes | Learning without a deadline or a fee |
| 8 | OER Commons | A discovery layer over many repositories; not a library itself but a good map of one | Aggregates thousands of resources | Mixed | Finding resources in subjects with thin open coverage |
| 9 | DOAJ | Peer-reviewed open-access journals and their metadata | ~19,000+ journals | Articles, PDFs | Finding free journals to submit to or read |
| 10 | PubMed Central | Biomedical and life-science articles that are free to read, plus preprints | ~9 million+ articles | HTML, PDF | Life sciences, medicine, public health |
| 11 | arXiv | Preprints across physics, maths, CS, stats, economics and more | ~2.4 million+ papers | Newest research; finding a supervisor's reading list | |
| 12 | LibriVox | Human-read public-domain audiobooks, produced by volunteer readers | ~10,000+ titles | MP3, ZIP | Long commutes; learning a language; rest |
1. Project Gutenberg — the plain-text shelf
Gutenberg is the oldest item on this list and still the most useful for one reason: it distributes plain text. No styling, no images, no DRM, no JavaScript. Every ebook can be downloaded as a single .txt file, which means it can be read by any program ever written, grepped in full, fed into a script, or opened on a device that cannot render HTML.
The catalogue is dominated by English literature and history, with substantial holdings in theology, philosophy, mathematics and children's writing. It also holds the occasional technical manual, because nineteenth-century engineers published them and they are now out of copyright.
Practical notes. Most titles have an EPUB, a Kindle-compatible file and plain text. Download counts are visible and are a decent proxy for usefulness. The donation request on each page is genuine and the project does need it, but nothing stops you.
A catalogue record tells you a book exists. A plain-text file tells you what it says.
2. Internet Archive and Open Library — the borrowing shelf
The Internet Archive is the largest of these collections by item count and the least uniform in what it holds, and it runs several distinct services under one roof. Confusing them is a common source of disappointment.
- Open Library is the catalogue and lending layer. Hundreds of thousands of titles have scanned pages and can be "borrowed" — one-hour or two-week loans with digital lending rights, subject to account limits.
- The Wayback Machine stores snapshots of websites going back to the late 1990s, which matters when a study needs the version of a site that existed in 2011.
- Special collections hold recorded oral histories, early computing magazines, government documents, software and video, and are frequently the only surviving copy of something.
What it holds is enormous and unevenly catalogued. Search quality is better for books than for ephemera, and the "borrowable" state is set per item by the rights holder, sometimes generously and sometimes not at all.
3. OpenStax — the first-year-course shelf
OpenStax publishes peer-reviewed textbooks: the books go through the editorial process of a commercial publisher, with external reviewers, and the publisher then pays for the editing, the artwork and the hosting, and gives the book away.
The strength is coverage of high-enrolment subjects with high textbook prices — marine biology, economics, sociology, psychology, chemistry, physics, statistics, business, nursing and precalculus are all here. The style is a hybrid of textbook and course pack: short sections, worked examples, check-your-understanding boxes, and end-of-chapter exercises with solutions.
Formats are unusually good. Every title is free as a PDF, readable in the browser, and available in Kindle and print-on-demand formats, so you can buy a physical copy for the price of printing. Errata are published openly, which is how open textbooks stay honest.
4. Open Textbook Library — the comparison shelf
This is a catalogue, not a publisher. It indexes open textbooks from OpenStax, BCcampus, LibreTexts, and dozens of smaller presses and individual authors, and it adds something the others do not: reviews and adoption information from faculty who have actually taught the book.
That matters when you are choosing between two candidates for a syllabus. The reviews typically mention what is thin, what is dated, and what worked in class. Read the negative ones; they carry the information.
5. BCcampus Open Textbook Collection — the humanities and social-science shelf
Run in British Columbia, this collection is organised by subject and places unusual weight on the humanities, social sciences and business. It also publishes an attribution file for each book, which sounds bureaucratic and is genuinely useful if you are compiling a bibliography and want the right credit line without hunting through the copyright page.
6. MIT OpenCourseWare — the syllabus shelf
MIT OCW holds complete course packages from a large fraction of MIT's catalogue, and increasingly from other universities too. A package usually includes lecture notes in PDF, assignments, exams with solutions, and sometimes video.
Coverage is strongest in engineering, mathematics, computer science, physics and economics. The signature of an OCW course is the assignment: a problem set that resembles a real course's, with a model solution attached. That is what makes OCW valuable as a self-study spine.
7. OpenLearn — the distance-learning shelf
OpenLearn is The Open University's free course platform. Its materials are written for independent learners with no tutor and no deadline, which is a different design brief from a university lecture course: explanatory notes, self-assessment questions and feedback-driven activities. Formats are a mix of PDFs, web pages and video.
8. OER Commons — the map
OER Commons is a discovery service with an embedded editor for sharing. It aggregates from many repositories, so you will find duplicates and the occasional dead landing page. Use it for coverage in subjects where open materials are genuinely thin — nursing, vocational trades, early-years education — and always follow a resource through to its host repository before relying on it.
9. DOAJ — the journal shelf
DOAJ is a directory of open-access journals, all peer-reviewed, with screening that checks licensing and editorial standards. It is where you go when you want to publish, or when a citation points at a journal you cannot otherwise access. Many listed journals charge no author fee; some do, and DOAJ flags that.
10. PubMed Central — the life-science shelf
The public-access half of PubMed. Life-science authors deposit accepted manuscripts here, so you can often read the accepted version of a paper years after it appeared in a paywalled journal. Coverage is strongest in medicine, molecular biology, epidemiology and public health, and it includes preprints.
11. arXiv — the preprint shelf
arXiv holds manuscripts before peer review, and the fields are unevenly distributed: dense in physics, mathematics, computer science and statistics, thin in the humanities, growing fast in economics. Its main use for most learners is finding a researcher's reading list: search the author and look for the field's canonical open reference on their page.
12. LibriVox — the listening shelf
Volunteer-read audiobooks of public-domain texts. The catalogue is mostly classic literature but includes public-domain science, philosophy and the occasional textbook-era work. Everything is plain MP3, free, with no account required.
id="how-to-tell-a-real-download-from-a-catalogue-record">How to tell a real download from a catalogue record
This is the most useful skill in the guide, and it takes ten minutes to learn.
A catalogue record is metadata: author, title, publisher, year, subject headings, an identifier. It tells you the work exists and where to look; it does not give you the text. A downloadable work is the text, the audio or the slides, in a file you can open. Search engines mix both and display them identically.
"Free to read" usually means free to read in the browser, if you have institutional access. "Free download" means the file is yours.
The six checks
-
Look for the file, not the "Read" button. A download link points to a file —
.pdf,.epub,.zip,.mp3,.html— and usually names the file and its size. A "Read online" button opens a viewer around a remote document and may still be gated by a login. -
Check what the download actually is. A "PDF" that is one page saying "purchase access" is a teaser. Open it and read the last page of the introduction. If the last chapter is missing, it is a sample.
-
Check whether an account or an institution is required. Institutional access is still paywalled if you are outside the institution. Many university libraries expose open-access content to the public as well; many do not.
-
Look for the licence. A licence statement tells you what you may do. No licence means all rights reserved, and downloading a copy for your own reading is a grey area even if it is technically reachable.
-
Look for a second copy. A book with an OpenStax PDF, a LibreTexts version and an Internet Archive scan is a real open book. A book with one link to a university library catalogue is a record.
-
Use the bibliographic record to confirm the edition. When a free PDF says "preliminary version, do not cite" or names a different edition from the one your course assigns, you have a mismatch. Check the ISBN or DOI in the record.
Download versus record: what each one gives you
| Signal | Real download | Catalogue record |
|---|---|---|
| Link target | Ends in .pdf/.epub/.zip/.mp3 | Ends in /item/, /record/, /search? |
| File size shown | Yes | No |
| Licence shown | CC BY, CC BY-SA, Public Domain | "All rights reserved" or nothing |
| Works offline | Yes | No — needs the site |
| Text searchable in your own tools | Yes | No |
| Citable without permission | Usually yes | Sometimes via a paywall-free route |
| Lasts if the site changes | Yes, you have the file | No |
| Typical use | Reading, study, teaching | Finding out where to read |
A three-minute habit
Before you build a reading list on a source, take three titles at random and try to open each one end to end. If you can reach the last chapter on all three, the source is real. If you cannot, the source is a discovery layer and you need to follow each item to its host before trusting it. This takes three minutes and saves you from building a term's worth of links that dead-end.
id="free-textbooks-subject-by-subject">Free textbooks, subject by subject
Open textbooks cluster where textbooks are expensive and enrolment is high. That is where the value is, so start there. The table below is a working core library: for each subject, one or two named open textbooks that are genuinely free to download, plus a note on what each covers.
| Subject | Open textbook | Publisher | Format | Notes |
|---|---|---|---|---|
| Biology | OpenStax Biology 2e | OpenStax | PDF, web, Kindle | Full first-year course; covers cell biology, genetics, ecology |
| Chemistry | OpenStax Chemistry 2e | OpenStax | PDF, web, Kindle | Atoms through electrochemistry; full worked problem sets |
| Physics | OpenStax College Physics | OpenStax | PDF, web | Algebra-based; pairs with The Physics Classroom for drill |
| Economics | OpenStax Principles of Economics 2e | OpenStax | PDF, web | Micro and macro; strong on the applied policy examples |
| Sociology | OpenStax Introduction to Sociology 3e | OpenStax | PDF, web | Includes research-methods coverage and a full ethics chapter |
| Psychology | OpenStax Psychology 2e | OpenStax | PDF, web | Covers the AP Psychology syllabus closely |
| Statistics | Introduction to Modern Statistics | OpenIntro | Free with a full R/Jupyter lab list; paid print only | |
| Programming (Python) | Python for Everybody | Charles Severance | PDF, web | No prior programming assumed; video companion |
| Programming (JS) | Eloquent JavaScript | Marijn Haverbeke | Web, PDF | Free to read online; project-based, no build tooling needed |
| Programming (Rust) | The Rust Book | Rust Foundation | Web, PDF | The de facto reference; requires some programming background |
| Git | Pro Git | Scott Chacon & Ben Straub | PDF, web | Free with GitHub-hosted version; the best version-control intro |
| Systems programming | Structure and Interpretation of Computer Programs | MIT Press | Web, PDF | The classic; the online edition is fully free |
| Computer systems | Computer Systems: A Programmer's Perspective | Pearson (author site) | Author's free PDF; the paid edition is separate | |
| TypeScript | The TypeScript Handbook | TypeScript team | Web, free PDF | Official, maintained, covers the current language |
| Automation | Automate the Boring Stuff with Python | Al Sweigart | Web, free PDF | Practical scripting for non-programmers; no prior code needed |
| Command line | The Linux Command Line | William Shotts | Free PDF from the author's site; a shell tutorial with exercises | |
| Political science | OpenStax American Government 3e | OpenStax | PDF, web | Constitution, parties, media, civil liberties |
| Business | OpenStax Principles of Management | OpenStax | PDF, web | Includes strategy, HR and organisational behaviour |
| Anatomy | OpenStax Anatomy and Physiology | OpenStax | PDF, web | System-by-system, with clinical links |
| Law | Constitutional Law: Principles and Policies | Erwin Chemerinsky | PDF (via publisher) | Advanced; check the publisher's free-textbook programme |
| Finance | Corporate Finance (Berk & DeMarzo) | Pearson | Partial free | The publisher runs a free chapters programme — check what is posted |
| Education | Foundations of Education | Wisconsin OER | Common-core aligned; good for early-years and secondary routes | |
| Nursing | Nursing Fundamentals | Open RN / BCcampus | Free, peer-reviewed; useful for pre-licensure revision |
How to read the table honestly
Statistics: Introduction to Modern Statistics (OpenIntro) is free in PDF, labs included. The print version is sold and is not materially different. You will need to install R, or use the browser-based labs.
Computer systems: Computer Systems: A Programmer's Perspective is a commercial Pearson title. The author posts the previous edition's PDF free on his own site for personal study. Read the edition carefully: it is a complete textbook, but not the current printing, so exercise numbering will not match a current course kit.
Law: Constitutional Law: Principles and Policies by Erwin Chemerinsky is a leading treatise. The publisher makes opening chapters free and runs open-textbook programmes for adoption. Verify what is posted before promising a whole course on it.
The two "textbook-like" sites worth bookmarking
The Physics Classroom is not a book; it is a free tutorial site with worked explanations, diagrams and practice problems on mechanics, electricity, waves, optics and light. It fills the gap a textbook chapter leaves when a concept has not clicked.
Seeing Theory is a set of free interactive visualisations of statistics — distributions, confidence intervals, hypothesis tests, regression — built with sliders and animations. It makes several statistical ideas obvious in ten minutes that a page of algebra does not.
Open Music Theory is a peer-reviewed, open-access textbook on music theory, published by a consortium including the University of Georgia and Seton Hall, covering notation, harmony, counterpoint and form with a large collection of sound examples.
Choosing between two candidates
- Check the table of contents against the course outline. Not the blurb — the actual chapter list. Two books titled the same thing can cover completely different ground.
- Check the exercises. A textbook without end-of-chapter problems is a reading, not a course. Open books vary wildly here.
- Check the media. Worked examples and video are the difference between "read it" and "understood it".
- Check the licence if you intend to remix: a CC BY-NC book cannot go in a paid course pack.
- Check the error status. OpenStax publishes errata openly; a self-published book may have silent errors. If the subject is one where an error matters — statistics, chemistry, law — prefer the maintained book.
- Check for a print option. Several open textbooks sell a print-on-demand copy for roughly the price of printing, and some institutions will pay for it from a study budget.
id="free-courses-and-course-materials">Free courses and course materials
A course is a better thing than a textbook in one specific way: it tells you the order.
The courseware comparison
| Platform | Courses | Includes | Video | Assessment | Best use |
|---|---|---|---|---|---|
| MIT OpenCourseWare | ~2,000+ | Lecture notes, assignments, exams, readings | Yes, many courses | Exams with solutions, no grading | Following a full syllabus unaided |
| OpenLearn (Open University) | ~300+ | Notes, activities, self-assessment quizzes, some video | Some | Quizzes with feedback | Guided self-study with structure |
| Saylor Academy | ~300+ | Full course pages, readings, discussion prompts | No | Assignments, no credit | Accounting and business pathways |
| Khan Academy | Thousands of short lessons | Video, practice problems, mastery system | Yes, throughout | Auto-graded exercises | Drilling the basics, gap-filling |
| LibreTexts | Thousands of pages | Peer-reviewed wiki textbooks, homework | No | Problem banks | One textbook spread across many books |
| NPTEL (India, NPTEL) | Thousands | Lecture videos + transcripts + books | Yes | Assignments via SWAYAM | Engineering and science at lecture level |
| Academic Freedom Project | Hundreds | Lecture videos with transcripts | Yes | No | Primary-source-driven US history and civics |
| Pressbooks | Thousands of books | Self-published textbooks with web/PDF/ebook export | No | No | Courses with a strong reading list |
MIT OCW and OpenLearn are complete course packages. Khan Academy and NPTEL are lecture collections — superb for drilling a specific topic, less good as a spine because they were never sequenced into a course. LibreTexts is a textbook platform, not a course, but its homework banks are extensive.
A textbook tells you what to read. A course tells you what to read, in what order, and what problem to attempt afterwards.
Saylor Academy
Saylor runs roughly three hundred free, self-paced courses, with unusual depth in accounting, business and the social sciences — free courses on financial accounting and corporate finance exist in full. No fee, no enrolment limit, no credit. Expect 15 to 25 hours per course.
Khan Academy
The right tool for a specific job: closing a gap you can name. "I do not understand partial fractions", "I do not know how to normalise a distribution" — Khan is close to unbeatable. The mastery system does not move on until you are right, and it tells you whether you are guessing or learning. Its weakness is depth: superb for the first two years of a subject, thin on third-year material, laboratory work and anything requiring an opinion.
NPTEL
The Indian government's National Programme on Technology Enhanced Learning provides lecture courses with video, transcripts and downloadable books, mostly in engineering, science and management, taught by professors from Indian institutes of technology. Many courses are free to watch; some certificates are paid. The transcripts make the videos searchable, which matters more than it sounds — you can find the ten minutes of a two-hour lecture that covers your problem.
LibreTexts
A wiki-based textbook platform with peer-reviewed pages. Its distinctive strength is homework: thousands of question banks with worked solutions, and a "Studio" feature that turns a page into a self-marking exercise. Many popular books exist here in parallel with a conventional PDF. Check the page history before you rely on a page — wikis get revised.
Academic Freedom Project
A US history and civics library of lecture videos with full transcripts, covering the American Revolution through the twentieth century, each anchored in primary documents. Transcripts mean you can read instead of watch, which is faster and searchable.
How to use a free course when you are not enrolled
- Get the syllabus, not the videos. Most courses publish a syllabus with the week-by-week plan. That plan is the course. Put it in your calendar before you watch anything.
- Watch at 1.5× or read the transcript. If transcripts exist, read. If not, the second viewing is where the learning happens; the first is for orientation.
- Do the assignments on paper, closed-book. The moment you look at a solution, the problem is gone. Write your answer first, then compare.
- Do the exam last. OCW exams are usually published with solutions. Take the exam under time pressure before you open the solution file, and mark your own work honestly.
- Write a one-page summary per week. Half a page of your own words on what you did not understand is worth more than the videos.
- Pick one anchor course and finish it. Finishing a single course end to end teaches more than sampling eight.
id="research-papers-and-academic-sources">Research papers and academic sources
The distinction between finding and reading research papers is where most people give up. Search tools index far more than they let you read.
Where the papers actually are
| Source | Coverage | What is free | Access model | Typical lag |
|---|---|---|---|---|
| PubMed Central | Biomedicine, life sciences | Accepted manuscripts and some final versions | Free to read | Often ahead of the journal issue |
| PubMed | Same index, plus paid journals | Abstracts only for most | Abstract free, full text via PMC | Current |
| arXiv | Physics, maths, CS, stats, econ | Preprints | Free to read | Days to weeks |
| DOAJ journals | All disciplines, ~19,000+ journals | Full text | Free to read | Current |
| Semantic Scholar / Crossref | All disciplines | Metadata and some full text | Metadata free | Current |
| Google Scholar | All disciplines | Links, some PDFs | Links only; often paywalled | Current |
| Zenodo | All disciplines | Accepted manuscripts and datasets | Free to read | On deposit |
| CORE | Aggregates repositories worldwide | Full texts where deposited | Free | Varies by source |
| Europe PMC | Biomedicine, wider than PubMed | Full text including preprints | Free | Current |
The preprints problem
Preprints are manuscripts released before peer review. That is their value and their risk. In fields where preprints are routine (physics, computer science, biology) a preprint is close to a normal research object; in fields where they are rare, treat them as work in progress.
The 2020 COVID-19 preprint episode taught the field a hard lesson: many high-profile preprints were wrong, and headlines amplified them. The lesson for a reader is simple — a preprint's claims are unreviewed, so reproduce or distrust, but do not cite as established.
A working literature-search procedure
- Start from a review article, not a primary paper. Reviews list the literature you need and are freely available in most fields. Find a recent, open-access review and read its reference list.
- Find the paper the field treats as canonical. Every field has two or three papers everyone cites. Ask a tutor, or search for the phrase "X is the standard reference for" — but better, find the paper's own "further reading".
- Check the open-access status of the exact DOI. A paper being indexed is not the same as being readable. Test the DOI before you put it on a list.
- Use the repository route. For paywalled papers, search the title on Zenodo, CORE, PubMed Central or the author's institutional repository. A large share of paywalled papers have a free green OA copy deposited by the author.
- Read the abstract, then the methods, then the results. In that order. Many people read the discussion first and end up reading the authors' opinion rather than the evidence.
- Read one paper per week, properly. Ten minutes with a good paper, annotating as you go, beats an hour of skimming five.
The goal is not to collect papers. It is to be able to explain, from memory, what a paper found and what it could not show.
Reading a paper properly: six passes
- Title and abstract. Does it answer a question you care about?
- Introduction. What gap does it claim to fill?
- Methods. Who was studied, how many, and how were they selected? Look for the sample size and the inclusion criteria.
- Results. Read the tables and figures before the prose. Authors write around their figures.
- Discussion. What limitations do they admit?
- References. Which earlier papers recur? Those are your next reads.
The evaluation questions that matter
- How many people, and how were they recruited?
- What was the comparison group?
- Was the outcome measured or self-reported?
- Was the paper's own limitation section naming the thing you suspected?
A paper is a tool, not a fact. Read it for what it measured, not for what you wanted it to say.
Datasets and reproducibility
Open data lives in Zenodo (a DOI is minted per deposit, which makes citation easy), Dryad, and discipline-specific archives. Cite the version of a dataset, not just the site. When you use someone else's code, look for a repository with a DOI and a licence, and record the commit or release you used.
id="audiobooks-podcasts-and-learning-by-ear">Audiobooks, podcasts and learning by ear
Reading is not always the fastest way in, and for some material it is not even the best. Anything with a narrative, a worked derivation or a spatial argument often lands better in the ear.
What is actually free to listen to
| Source | Holdings | Length / format | Cost | Best for |
|---|---|---|---|---|
| LibriVox | ~10,000+ volunteer-read public-domain audiobooks | 1–40 hours, MP3 | Free, no account | Classic literature, public-domain science |
| Internet Archive audio | Recorded lectures, radio archives, oral histories, audiobooks | Varies, MP3 | Free | Archive material and older broadcast |
| Project Gutenberg audio | Audible-read public-domain texts | Varies | Free | Library-style classics |
| Public domain films (Internet Archive) | Feature films, documentaries, scientific films | Full length, video | Free | Learning from documentary film |
| Podcast RSS feeds | Thousands of academic and professional podcasts | 30–90 min/episode | Free | Current thinking, professional practice |
| OpenLearn / OCW video with audio | Course lectures | 10–60 min | Free | Studying by ear during travel |
LibriVox
LibriVox is the clearest win in this section. Volunteers read public-domain texts aloud and the recordings are released under a public-domain dedication. The catalogue is dominated by classic literature but also holds public-domain science and mathematics, which is a pleasant surprise: nineteenth-century popular science books were written to be read aloud to a family, and they still work.
Practical points. Everything is plain MP3, in single files or zipped by chapter. There is no account, no app required and no listening limit. Most titles have a text edition on Project Gutenberg or Gutenberg's audio partner, so you can read along if you want the words. Length varies from a two-hour short story to a forty-hour novel; check the total before you commit, and use the chapter files to sample the narrator's pace.
What podcasts are good for
A podcast is a lecture with the visual aids removed. It is excellent for surveying a field you know nothing about, for hearing working researchers talk about their own work in more detail than a review article allows, for staying current in a fast-moving area, and for deciding whether a topic deserves three hours of reading.
It is bad for anything needing precise notation, equations you must re-read, or structured revision before an exam. You cannot skim a podcast or look up a definition mid-sentence. Treat a podcast as a first pass and a book as the second.
A listening routine that works
- Choose one audiobook and one lecture series. Two audio streams, not five.
- Keep a single playback rule: anything longer than 15 minutes goes to a commute or a walk, never to a desk.
- Take notes by hand after each session, three lines only — what it was about, one idea worth keeping, one question. Handwritten notes are retained noticeably better than typed ones for material heard once.
- Follow up within 48 hours. Write down the one term you did not know and look it up. Audio without follow-up leaks away within a week.
- Do not listen and read the same thing at once. Two channels, one brain. If a piece is important, pick one mode.
id="public-domain-books-for-the-general-reader">Public-domain books for the general reader
If your goal is reading rather than studying, the shelf is enormous and slightly strange. It is dominated by books that are free because their authors died long ago, and by a smaller set of books that are free because their authors gave them away.
What the public domain actually contains
Roughly, the public domain in English consists of everything published before 1930 (in the US as of 2025, subject to a complex renewal rule), a large body of 1930–1963 works whose copyright was never renewed, and works published after 1963 whose authors deliberately dedicated them to the public domain — most famously The Little Prince by Antoine de Saint-Exupéry, and The Story of Ferdinand by Munro Leaf.
That last group is the interesting one. A handful of mid-century children's books and a few well-known works were dedicated to the public domain by their authors or their estates, and because permission was granted, they can be freely distributed. Do not assume a 1940s book is free; many are not.
A reliable starter set
| Work | Author | Why it is worth your time | Where |
|---|---|---|---|
| The Adventures of Sherlock Holmes | Arthur Conan Doyle | Twelve tight stories; the template for modern detective fiction | Project Gutenberg |
| Pride and Prejudice | Jane Austen | Sharp social observation; the dialogue carries the characterisation | Gutenberg, LibriVox |
| Frankenstein | Mary Shelley | The first science-fiction novel; the 1818 introduction frames the ethics | Gutenberg, LibriVox |
| Middlemarch | George Eliot | Often called the best novel in English; long, worth it | Gutenberg |
| Meditations | Marcus Aurelius | Stoic practical philosophy; short sections for daily reading | Gutenberg |
| The Critique of Pure Reason | Immanuel Kant | Dense; use the Gutenberg HTML or a good translation with notes | Gutenberg |
| On the Origin of Species | Charles Darwin | Still readable, still argued about | Gutenberg |
| The Souls of Black Folk | W. E. B. Du Bois | Foundational; written in a spare, beautiful prose | Gutenberg, LibriVox |
| The Autobiography of Malcolm X | Alex Haley | A major twentieth-century text in the US public domain | Internet Archive |
| The Little Prince | Antoine de Saint-Exupéry | Dedicated to the public domain by the author | Gutenberg |
| A Room with a View | E. M. Forster | Shorter than the others; the Italy chapters are the point | Gutenberg |
| The Secret Garden | Frances Hodgson Burnett | Dedicated to the public domain; a genuine comfort read | Gutenberg, LibriVox |
For non-fiction, the public domain holds a great deal of popular science, mathematics and engineering: Charles Darwin, Alfred Russel Wallace, the works of George Bernard Shaw, and the mathematical treatises of Euler and Euclid, many of them in translation. Euclid's Elements is available in a free Dover edition and in a cleaned-up online version.
Why the public domain is not automatically good
Age is not quality. A large share of the shelf is dull nineteenth-century fiction written to conventions that no longer hold; sample before you commit. Metadata is inconsistent, so older Gutenberg files may lack a table of contents or carry OCR errors in poetry — use the HTML or EPUB version for anything with unusual formatting. And classics come with editors who had opinions: for anything you will cite, check the edition against a scholarly one, or use the free copy for reading and cite the standard edition.
What is not free, and why that matters
Most books from 1930 onward are still in copyright, and their authors and estates are still paid. The right conclusion is not that you should hunt for questionable copies; it is that the shelf is genuinely large and legally clean without them. And for the in-copyright books you want, there are two honest routes: the library, and the borrowable copy on the Internet Archive.
id="building-a-reading-list-that-gets-finished">Building a reading list that gets finished
Most people do not fail at reading because the books were bad. They fail because the list was too long, too vague, or too disconnected from anything they had to do.
The arithmetic of finishing
A realistic evening is 30 to 60 minutes of focused reading. At roughly 230 words per minute for non-fiction prose, that is 7,000 to 14,000 words a day — 30 to 60 printed pages.
Work from that. A 12-week plan with four reading sessions a week gives you roughly 200 sessions. At 40 pages a session, that is 8,000 pages. A 300-page book is about 7.5 sessions. So a realistic twelve-week commitment is roughly twelve to fifteen books of average length, not forty.
If your list does not fit in a year of evenings at 40 pages, it is a wish list, not a reading list.
The three-list system
Keep three separate lists, never mixed.
- The current book. Exactly one, on your desk, in your bag, or loaded on your phone.
- The next three. A short queue, so you start the next one without deciding.
- The someday list. Unlimited, unranked, guilt-free. A someday list is a gift to your future self, not a debt.
The commonest cause of a stalled list is the absence of a "current book" slot — a commitment to one thing at a time.
Choosing what goes on the list
Use a three-source mix, so the list has both depth and breadth.
| Source | Proportion | Why |
|---|---|---|
| One authoritative book on the subject | 40% | Gives you the vocabulary and the shape of the field |
| Three to five open textbooks and course notes | 30% | Structured, tested, with exercises |
| Two or three public-domain or classic works | 20% | Context, and the pleasure of reading something older |
| Two or three recent papers or essays | 10% | Keeps you current; sharpens the other ninety per cent |
A list made only of textbooks is a syllabus, not a reading list. A list made only of papers is unreadable.
A procedure for building any reading list
- Write the one-sentence question the list answers. "How do I understand plate tectonics well enough to teach it?" is a list. "Geology stuff" is not.
- Choose one anchor. The book that will carry the vocabulary. Read it first, because it gives you the terms to search the rest with.
- Add one open textbook per sub-area, chosen by its table of contents, not its cover.
- Add one course. A free courseware package gives you the sequence and the assignments that a book does not.
- Add two to three classics or public-domain works that people in the field actually read.
- Add three recent papers from the last two years, chosen via a review article.
- Add one practical artefact — a dataset, a code repository, an atlas, a case study. Reading with a tool attached teaches more.
- Order it. A reading list without an order is a pile. Write the order in.
- Cut it to 40% of its current length. Always. You can add back later.
The weekly rhythm
- Monday — the anchor chapter. The hardest, most important reading, done when you are fresh.
- Tuesday — the open textbook section plus its exercises. Exercises, not just reading.
- Wednesday — a paper, read in the six-pass way. Short and structured.
- Thursday — the classic or the audio equivalent. The easy one. It keeps the chain alive on a bad week.
- Friday — review and update. Twenty minutes: write what you learned, update notes, queue the next book.
On a bad week, do only Thursday. The habit matters more than the volume, and a four-page week beats a missed week.
Notes that work
- A question note. One question per session, in your own words — these become your revision sheet.
- A concept note. A definition, a diagram you redraw, or a formula with a sentence on what it is for.
- A quote note. Three to five per book, with page references.
Do not highlight. Highlighting feels productive and retrieves almost nothing. Writing a sentence in your own words is slow enough to be useful and fast enough to be repeatable.
id="a-subject-by-subject-starter-list">A subject-by-subject starter list
This is the shelf to start from, one subject at a time. Each row gives a core open resource, one free companion for depth, and what to do with them. Use it as the starting point for a reading list, not as a finished syllabus.
STEM
| Subject | Core open resource | Companion | Do this |
|---|---|---|---|
| Mathematics | Introduction to Modern Statistics (OpenIntro) for stats; for pure maths, Calculus by Gilbert Strang (free PDF) | MIT OCW 18.01 Single Variable Calculus | Work every problem; Strang's lectures are the free top-up |
| Physics | OpenStax College Physics | The Physics Classroom for drill; MIT OCW 8.01 | Redraw each diagram yourself; do the practice questions before the exercises |
| Chemistry | OpenStax Chemistry 2e | MIT OCW 5.111 / 5.111 principles | Focus on stoichiometry and equilibrium; draw the reaction mechanisms |
| Biology | OpenStax Biology 2e | MIT OCW 7.01 (Genetics) | Draw the processes rather than memorising them |
| Computer science | Structure and Interpretation of Computer Programs; Eloquent JavaScript; Python for Everybody | MIT OCW 6.001; 6.0001 | Do the exercises; read the code, then change it |
| Statistics / data | Introduction to Modern Statistics | Seeing Theory for the visual intuition | Run the R labs; then re-derive one estimator by hand |
| Engineering | MIT OCW 2.001, 2.002, 2.003 | Course notes + problem sets | Treat problem sets as the course; exams with solutions as the test |
| Astronomy | Astronomy (OpenStax) | NASA open data and the Hubble and JWST archives | Look up real objects; do not only read about them |
| Earth science | Open textbook options via Open Textbook Library | MIT OCW 12.001 | Work the problems; map each process to a landscape |
Social sciences and humanities
| Subject | Core open resource | Companion | Do this |
|---|---|---|---|
| Economics | OpenStax Principles of Economics 2e | CORE Econ's The Economy (open, online, with interactive tools) | Read one chapter of each and compare the styles |
| Sociology | OpenStax Introduction to Sociology 3e | OpenStax Sociology 3e methods chapters | Write a one-page observation report on a public space |
| Psychology | OpenStax Psychology 2e | MIT OCW 9.00SC | Design one small study from the research methods unit |
| Political science | OpenStax American Government 3e | Constitution Annotated (Congressional Research Service) | Read the primary documents alongside the summary |
| History | The Souls of Black Folk; MIT OCW 21M.011 | Academic Freedom Project | Read one primary source per chapter |
| Philosophy | Meditations (Gutenberg); The World's Religions (Global Religions Library) | Introducing Philosophy (OpenStax) | Write your own position on one question |
| World literature | Frankenstein, Middlemarch, The Little Prince (Gutenberg) | LibriVox audio of the same titles | Read one; listen to one |
| Education | Foundations of Education (Wisconsin OER) | OpenStax Introduction to Sociology for context | Write a lesson plan using only open materials |
| Business | OpenStax Principles of Management | Saylor Academy's free accounting course | Build a simple budget from the book, not from a template |
| Law | OpenStax Business Law; Constitutional Law: Principles and Policies free chapters | Cornell Legal Information Institute | Read the actual statute text, not only the summary |
| Communication | Say No to the Fake News / MIT OCW 21W.789 | OpenLearn's free communication courses | Fact-check three articles you have shared this week |
| Gender studies | OpenStax Introduction to Sociology gender chapter | MIT OCW 21A.103 | Read two opposing scholarly pieces |
Health, education and professional preparation
| Subject | Core open resource | Companion | Do this |
|---|---|---|---|
| Nursing | Nursing Fundamentals (Open RN) | BCcampus open textbooks | Do the clinical calculations chapter cold |
| Medicine / public health | PubMed Central open-access reviews | WHO and CDC open resources | Write a 500-word plain-language summary of one study |
| Medicine / clinical | StatPearls (NCBI Bookshelf, free) | NLM open resources | Use it as a reference, but check the date on every entry |
| Teacher preparation | Foundations of Education | OpenStax Psychology chapters on development | Plan a differentiated lesson |
| Social work | Human Behaviour and the Social Environment (open textbook) | OpenStax Sociology | Case-read one chapter per week |
| Business / accounting | Saylor Academy accounting courses | OpenStax Principles of Accounting | Close the books on a small dataset of your own |
Language and the arts
| Subject | Core open resource | Companion | Do this |
|---|---|---|---|
| Spanish / French / German | Project Gutenberg's bilingual primers and parallel texts | LibriVox in the target language | Read the Gutenberg text while listening to the audiobook |
| English literature | Gutenberg classics; The Souls of Black Folk | Project Gutenberg's periodicals (Shakespeare's The Tempest, 1911) | Read one play aloud with two voices |
| Music theory | Open Music Theory | Society for American Music theory resources | Play the examples; do not only read about them |
| Film studies | Public-domain feature films (Internet Archive) | Early film criticism in periodicals | Watch one film with no subtitles, then with them |
| Creative writing | Public-domain short stories; The Yellow Wallpaper (Gutenberg) | Manuscript and archive collections | Write one page a day for a fortnight |
A good starter list is not the longest one. It is the one where every entry has a named source, a named format and a date you could start.
id="twelve-week-reading-plans">Twelve-week reading plans
Three plans, depending on how much time you have. Each is designed to finish inside twelve weeks without slipping.
Plan A — the steady reader (three sessions a week, ~40 pages each)
| Week | Focus | Sources | Output |
|---|---|---|---|
| 1 | Orientation | One anchor book's intro and contents; set up the three-list system | Written one-sentence question; notebook open |
| 2 | Anchor, part 1 | Anchor book, sessions 1–3 | 3 question notes |
| 3 | Anchor, part 2 | Anchor book, sessions 4–6 | 3 concept notes; 1 quote note |
| 4 | Anchor, part 3 + first classic | Anchor sessions 7–9; one short classic | 3 more notes; classic read in one sitting |
| 5 | Open textbook | One subject textbook chapter plus exercises | Exercises attempted before solutions |
| 6 | Paper week | One open-access paper, six-pass read | 200-word summary |
| 7 | Second classic | Second classic, plus audiobook of the first | 3 notes; one listening session |
| 8 | Open textbook, harder part | Textbook chapter with problem sets | Full problem set, marked |
| 9 | Course week | One OCW or OpenLearn module, with its assignment | Assignment completed closed-book |
| 10 | Practical | Dataset, code, map or atlas for the subject | One thing produced, not just read |
| 11 | Synthesis | Re-read your notes; write a 500-word synthesis | 500 words, unedited |
| 12 | Review and plan | Write a 300-word plan for the next twelve weeks | Plan dated and in the calendar |
Plan B — the exam candidate (six sessions a week, targeted)
Twelve weeks of a course syllabus, worked in the order that suits the exam.
- Week 1 — map the syllabus. Turn every topic into a page number and a chapter. Colour-code by confidence: green, amber, red.
- Weeks 2–7 — one week per topic. Red topics first, because they need the most time. The pattern for each week: read the textbook section, do the exercises, then watch the matching course lecture.
- Weeks 8–9 — past papers. Do three past papers under exam timing, marking them with the official scheme yourself.
- Week 10 — error log. Every mistake becomes a line in a log: question type, what you misread, what you should have done. This log is worth more than the past papers.
- Week 11 — the red topics again. Redo the red ones without notes. Then redo the amber ones.
- Week 12 — mock and rest. One full mock, then a light review of the error log only. Do not start new material in the final week.
Plan C — the researcher (writing fortnightly, ~10 pages a week)
- Fortnights 1–2 — survey and narrow. Read one open-access review plus three more; build an annotated bibliography of twelve items, select five, and write a two-page framing note.
- Fortnights 3–5 — read, method, data. Five papers read in six passes each (1,500 words of notes); one method reconstructed in plain language; the underlying dataset or code located and cited.
- Fortnights 6–9 — draft. Write the framing, introduction and body, then revise the introduction against two papers that contradict your framing, and follow those to ten new references.
- Fortnights 10–12 — feedback and file. Give one paper to a colleague and take one in return; keep a feedback log; finish with a pass on citations, licence and deposit. A public draft is a real output.
The "if you only have 30 minutes" table
For weeks when time collapses. Keep this visible.
| If you have | Do this | Why it works |
|---|---|---|
| 30 minutes a day | One chapter of the anchor book, 40 pages, no phone | Continuity beats volume |
| 30 minutes a week | One page of notes on where you got to, plus one question | Prevents restart friction |
| 30 minutes a month | One podcast episode on the subject | Keeps the thread alive |
| 30 minutes total | One audiobook chapter while the kettle boils | Habit token, not real study |
| 30 free minutes plus a library | Any borrowed book you have been meaning to start | Removes the only barrier |
| 30 minutes and no signal | A Gutenberg plain-text file on your phone, offline | Works with no connection |
| 30 minutes and a tired brain | One classic, or one audiobook | Low-effort entry keeps the chain |
The plan fails in the week you have no time. Design for a 30-minute version of the plan, or you have no plan.
id="using-these-with-a-library-a-course-or-a-study-group">Using these with a library, a course or a study group
With a library
The library's job changes when the open shelf exists. It is no longer mainly a place to get a copy of a book. It is the place to get the copy you cannot download, plus trained staff, plus physical space.
- Ask for the in-copyright item locally. If a book is not on the open shelf because it is in copyright, the library may have it in print or as a licensed e-book.
- Ask for the Internet Archive copy. Many libraries hold Internet Archive partner-lending access, giving you borrowable digital copies of in-copyright books. Staff can usually set this up in ten minutes.
- Use interlibrary loan for the hard-to-get. Standard ILL can fetch a physical copy of almost any published book within one to three weeks, and electronic ILL often delivers a scan within days.
- Use the librarian for the reading list itself. A subject librarian can point you at the standard open textbook for your course faster than any search engine.
- Ask what is on reserve. Reserve lists increasingly include open textbooks, and they tell you exactly what the course expects.
With a course
- Ask the lecturer for the open alternative. Many now teach from an open textbook. If they use a commercial one, ask whether the library has an access key for students, and whether an open alternative exists.
- Ask for the syllabus as a document. Course syllabi with reading lists are often public, and the reading list gives you the vocabulary to search for.
- Replace one item, not the whole list. Find a free equivalent for one expensive textbook, then add one free supplement. Two changes are a start; a wholesale replacement is a negotiation nobody wants.
- Watch the courseware. If the course is on OCW or OpenLearn, take it alongside the live course. The assignments are the same.
- Use the error log, not the textbook's solutions. The free textbook gives you solutions; the course gives you the exam weighting. You need both.
With a study group
- Divide the reading rather than the reading aloud. Each person takes a different chapter of the anchor and produces a one-page summary with two questions. Five members, five pages, one session.
- Run the exam as a group. Sit the paper individually for 40 minutes, mark in silence, then compare marks and reasoning.
- Use the shared error log. One document, one column per person, one row per mistake. Nobody repeats another person's mistake; the group gets faster fast.
- Alternate the medium. One week of reading, one week of listening, one week of a video course. Mixed formats expose gaps that a single format hides.
- Keep the group small and the commitment public. Four people who will show up beat nine who might.
- Set a hard end date. Twelve weeks, then stop. Reading lists without an end become an identity rather than a plan.
A note on citing open material properly
Most open material needs a citation like any other source: author or institution, title, publisher or repository, year, DOI or permanent URL, and the date you accessed it. Record the version — OpenStax books carry edition numbers and LibreTexts pages carry revision dates. And if a licence requires attribution, a CC BY book wants "Title by Author, licensed under CC BY 4.0, via Source" plus a link.
id="common-mistakes-and-how-to-get-out-of-them">Common mistakes and how to get out of them
1. Treating a search result as a book. The most common failure, and the reason a guide like this one has to exist. Search engines flatten a bibliographic record and a downloadable file into the same blue line. Get out of it: read the source's own description, and try to open one title end to end before trusting it. If you cannot reach the last chapter, it is a record.
2. Downloading a sample and calling it a book. Many catalogue pages offer the first forty pages. That is not the book. Check the last page of the introduction before you plan around a title. For a partial free edition, find the full edition on a different shelf rather than searching for an unauthorised copy.
3. Ignoring the licence. A book marked "for personal use only" or "no redistribution" is fine for you and useless for a study group or a course pack. If you plan to share, check the licence first. CC BY is the safe default for remix and redistribution.
4. Building a list of fifty books. Almost everyone does this and almost everyone abandons it by week three. Get out of it: cut the list to 40% of its length, then pick one as the current book. Everything else moves to the someday list, where it costs nothing.
5. Reading instead of doing. An open textbook with no exercises is a long document. Do the problems. The problem set is where the learning is, and it is the part everyone skips.
6. Confusing the free course with the free courseware. MIT OCW and OpenLearn give you the materials. They do not give you the teaching, the feedback, or the credit. If you need the credit or the marking, check whether the institution runs an accredited route to it. Several universities now offer free enrolment and free credit for OpenLearn courses, and a handful of universities recognise OCW course work toward a degree.
7. Reading only one format. The same book as text, as audio and as video is not redundant. Audio suits commute and tired brains; text suits precision; video suits physical and spatial ideas.
8. Storing books and never opening them. A download folder with 400 PDFs is a guilt archive, not a library. Keep a "to read" file with only the next three, and clear the rest from your desktop without deleting the links.
9. Quoting from a book you have not read. Open PDFs are easy to search, and it is tempting to pull a quotation from a book you never opened. If you cannot say what the argument around the quote was, do not use it.
10. Treating "free" as "cheap". Free materials are often better designed than expensive ones, because their authors are teachers first and because the licence requires them to be usable. Do not assume a paywalled book is better.
id="final-checklist">Final checklist
Before you start, work through this. It takes about fifteen minutes.
Scope
- [ ] I have written my reading list's one-sentence question.
- [ ] I know how many evenings a week I actually have, honestly.
- [ ] I have picked one anchor source.
Sources
- [ ] I have opened one title from each of my sources end to end.
- [ ] I have confirmed each file is a full work, not a sample.
- [ ] I have noted the licence for anything I plan to share or adapt.
- [ ] I have found the permalink, DOI or repository URL, not the search result.
Format
- [ ] Each item has a format I can actually use: PDF, EPUB, HTML, MP3.
- [ ] I have checked that anything I need offline can be downloaded.
- [ ] I have a note of the edition or version for each item.
Plan
- [ ] There is exactly one current book, and the next three are queued.
- [ ] The list has an order and a start date.
- [ ] There is a 30-minute version of the plan for bad weeks.
Practice
- [ ] I have booked three sessions in the calendar in the next seven days.
- [ ] I have a notebook or a file for question and concept notes.
- [ ] I know whether each item needs exercises or an assignment, not just reading.
- [ ] I have told one other person what I am doing, with a date to report back.
Ethics and honesty
- [ ] Everything I have is free to read and free to download, or I have marked the ones that are only free to read.
- [ ] I have not downloaded anything from a site that looks like an unauthorised copy.
- [ ] My notes distinguish between what a source says and what I think it means.
And one last thing. Everything in this guide is findable again, but finding it again is not the point. The point is that you now know which shelf holds what, how to check whether a link is a real file or a catalogue record, and how to turn any of it into a plan that survives contact with a busy week.
For a live search across these open libraries and more — checking the actual source tier for each title as you go — use anyfreebook.com. It searches the collections described above and tells you plainly whether a title is a verified public-domain download, a borrowable copy, or a catalogue record only.
Free is not a consolation prize here. It is the better shelf.