All Frontier Global · Free · No Login
An essay is not typing and a thesis is not length. Both are an argument you can defend and sources you can prove. Here you shape the argument, structure the long work, and check every source you cite — the three jobs of one late evening, on one page. Nothing here writes it for you; that is the whole point.
01 — One Sitting
Nobody writes the chapter, then hunts the references, then learns paragraph structure, in that tidy order. At eleven at night it is all one job: this sentence needs a sharper claim, the claim needs a source, and the source needs checking before it goes anywhere near the bibliography. So all three live together here.
Free, no login, no upload — and no version of this site will ever write the thing for you.The claim someone could argue with, the paragraph that earns it, the quotation that is doing work rather than filling space. Essay craft, in six moves.
Eighty thousand words is a different animal from eight hundred. The question, the literature map, the methodology, the chapter spine — and the viva at the end of it.
Paste a DOI: who wrote it, where it appeared, whether it was ever retracted, whether a legal free copy exists, and the citation formatted to hand. The unglamorous half of scholarship.
02 — The Essay
Marks are lost in the same six places every year, and none of them is vocabulary. Each card below is one move: what it is, why it fails, and a specimen edit — the weaker sentence struck through, the stronger one beside it. Steal the pattern, not the sentence.
Specimen edits use placeholder subjects on purpose. The shape is the lesson; the content has to be yours.A thesis statement is a sentence a reasonable person could disagree with. If nobody could take the other side, you have written a topic rather than a thesis — and the essay will drift into description for two thousand words before anyone notices.
Specimen editThis essay will discuss the poem’s imagery.The poem’s imagery is not decoration: it carries the argument the poem never states.
One idea per paragraph, four moves inside it: the claim, the evidence, the warrant — why this evidence supports this claim — and the link onward. The warrant is the move most often missing, and it is exactly the difference between quoting and arguing.
Specimen editClaim. Quotation. New paragraph.Claim. Quotation. Why the quotation proves it. What follows from that.
Quote when the exact words matter, paraphrase when the idea matters, summarise when only the shape matters — and attribute all three. The sentence that earns marks is never the quotation; it is the one immediately after it, in your voice, saying what the quotation is doing here.
Specimen editThree quotations, stacked, no commentary.One quotation, then two sentences of your own.
A first draft exists to find out what you think, and it is allowed to be bad. Write it fast, in whatever order is easiest, and leave square brackets where the evidence will go. Editing a bad page is work; editing a blank page is not a thing that can be done.
Specimen editWait until the argument is clear, then write.Write until the argument is clear, then rewrite.
Cut a tenth. Turn nouns back into verbs, keep the subject beside its verb, and read the thing aloud — the sentence you stumble over is the sentence your reader will stumble over, and it is almost always doing two jobs at once. Give it two sentences.
Specimen editconducted an investigation intoinvestigated
Description in an argument’s coat. Quotation stacked where reasoning belongs. A thesis nobody could dispute. A conclusion that introduces new material. A paragraph carrying two ideas. Each has the same cure: name the claim, then earn it in the next four sentences.
Specimen editIn conclusion, as we have seen, there are many factors.So the second reading holds — and it changes how the last stanza has to be read.
03 — The Thesis
A dissertation fails for structural reasons far more often than for stylistic ones: a question too big to answer, a literature review that lists instead of maps, a methodology written from memory eighteen months late. Eight rows, in the order they will actually come at you.
Doctoral, master’s or undergraduate dissertation — the spine is the same; only the scale changes.Eight structural decisions · make them early, revisit them monthly
A thesis question is one you can actually answer with the time, the access and the word count you have. Write it as a single sentence ending in a question mark, and if it needs a semicolon to hold together, it is two questions — pick one.
A map, not a list. Group by argument rather than by author: here is the settled ground, here are the two camps, here is the corner nobody has looked at — which is the corner you are standing in. If a paragraph begins with a surname, it is probably a list entry.
Full guide: literature review →What you did, and why it was the right thing to do — including what you chose not to do. Write it while you are doing it, in a running file. Reconstructed methodology sections read exactly like reconstructed methodology sections.
Each chapter answers one question, with its own evidence, and changes what the reader believes. The test: one sentence per chapter, read in sequence, should sound like a coherent argument. If two sentences could swap places without anyone noticing, the spine is loose.
Choose the style once, keep everything in a reference manager, and record the identifier — the DOI, not the download link — on the day you read the thing. Every hour you save now costs three in the final fortnight, always at the worst possible moment.
Somewhere in the middle the thesis quietly becomes about something slightly different. That is normal and often an improvement. When the introduction and the conclusion disagree, the conclusion is usually right — which is why the introduction is written last.
You are the world expert on your particular corner, and the room knows it. Prepare four things: the one-minute summary, the weakest chapter and why it is weak, what you would do differently with another year, and the question you hope nobody asks. They will ask it.
No new reading, no new arguments. Read the whole thing aloud in order, fix the joins between chapters, and check that every citation in the list resolves to a real, current record — correct authors, correct year, no retraction you never heard about. That last check is what section 04 is for.
04 — The Source Check
A citation is a promise that the thing exists, says what you say it says, and has not been withdrawn since you saved the PDF. Six fields keep that promise, and every one of them comes from public scholarly infrastructure that anyone can query. Here is what each field is, and why it matters more than it sounds.
Read this alongside the desk at the top of the page: same six fields, same sources.Authors, journal or archive, publisher and date, taken from the registry that issued the identifier: Crossref for most journal articles, DataCite for a great many preprints, datasets, software and theses. Two registries, because neither holds everything — a check that asks only one will confidently tell you a real paper does not exist.
The check almost nobody runs. Papers get corrected, and a few get retracted — and the PDF sitting in your downloads folder never updates itself to say so. Publishers file their retraction and correction notices with Crossref, which since September 2023 also holds the Retraction Watch database as a public resource.
Unpaywall answers one question well: has the publisher or the author made a version of this openly available — a repository copy, an accepted manuscript, an open journal? Often yes. Sometimes no, and then the honest answer is the publisher’s page and a walk to your library.
Journal article, preprint, chapter, dataset, thesis, software. It matters twice over: a preprint has not been through peer review, and a preprint gets revised — the version you read last year may not be the version on the server today. Say which one you read, and when.
Citation counts and the web of works around a paper, from the open scholarly index. Read them as engagement, never as proof of quality — a heavily cited paper can be heavily cited because it is disputed. The real value is finding the reply, the review and the correction that came after it.
The same metadata, arranged into APA, MLA or Chicago and ready to paste — then read once by you against your department’s own guide. House styles differ in small, marked ways, and the guide, not any tool, is what you are assessed against.
We read the parts publishers put out openly — the metadata and the abstract — summarise them in plain English, and then link you onward to the publisher’s page or to a legal open-access copy. We do not reproduce paywalled full text, and we do not route around a paywall. If there is no free copy, the check says so; that is a result, not a failure.
A worked instance, because it is the honest case rather than the flattering one: the 1953 Watson and Crick paper in Nature (10.1038/171737a0) is one of the most famous articles in science, and Unpaywall reports no open-access copy of it at all. So a check on that DOI ends with the record, the summary, the citation, and a link to the publisher. Your university library is the next step, and we will say so.
Sources — Crossref, “News: Crossref and Retraction Watch”, 12 September 2023, doi.org/10.13003/c23rw1d9. Open-access status of 10.1038/171737a0 read from the Unpaywall API on 25 July 2026 (reported: closed). Record details for the two example identifiers on this page read from the Crossref and DataCite APIs on the same date.
05 — The Shelf
None of this is our data, and that is the point — it is public scholarly infrastructure, open to anyone with a browser and a question. Here is exactly what we ask and whom we ask, so you can go straight to the source and check our working, or build something better with it.
No accounts, no scraping. What these services ask of us in return is usually just an email address, so they can tell polite traffic from rude.Six public services · free to query, free to check us against
The registry behind most journal-article DOIs: titles, authors, journals, dates — and the correction and retraction notices publishers file, alongside the Retraction Watch data it now holds openly. This is where a check starts.
crossref.orgThe other great registry: preprints, datasets, software, theses and reports, with their versions and their dates. Ask Crossref alone about a preprint identifier and you will be told, wrongly, that there is nothing there.
datacite.orgOne question, answered well: is there a legal open-access version of this work, and where does it live? It asks for an email address with the request rather than a key — and it will happily tell you the answer is no.
unpaywall.orgAn open index of works, authors, institutions and the citation links between them — where “what happened to this paper afterwards” comes from, including the replies and reviews that a plain search will bury.
openalex.orgThe open archive for physics, mathematics, computer science, quantitative biology and more, with a public interface over the whole of it. Where a great many preprint identifiers ultimately point.
arxiv.orgThe biomedical and life-sciences literature index, searchable through free public interfaces — the right first stop for anything clinical, where the version of record and its corrections matter most of all.
pubmed.ncbi.nlm.nih.govThe Line
There is no order form on this site, no upload box, no “send us the brief”. Everything here exists to help you understand your material, structure your own argument and verify your own sources — because the skill is the point, and because work you did not do cannot be defended in a seminar, in a viva, or at three in the morning when you are wondering whether you actually believe it.
Your institution sets the rules on what help is allowed, and those rules differ by university, by department and sometimes by module. Read yours, follow them, and when you are unsure, ask your supervisor before you write — not after. Declaring what you used costs you nothing and settles the question permanently.