How to read an element-by-element claim chart
What each column of a claim chart tells you, how to judge the evidence in each row, and what a partial finding means before you rely on one.
By the One Transform team7 min read
A claim chart sets out a patent claim one limitation per row and places beside each limitation the evidence that a product or reference meets it, a finding, and the reasoning that connects the two. Read it row by row, not as a whole. A claim is met only when every one of its limitations is met, so a chart is only as strong as its weakest row.
What a claim chart is for
The same format serves two opposite purposes. An infringement chart maps a patent's claims onto an accused product to show that someone, without authority, makes, uses, offers to sell, sells or imports the patented invention during its term, which is direct infringement under 35 U.S.C. § 271(a). An invalidity chart maps the same claims onto prior art to show that they were not new or were obvious.
Several district courts require charts by local rule. In the Northern District of California, Patent Local Rule 3-1(c) requires infringement contentions to include "a chart identifying specifically where and how each limitation of each asserted claim is found within each Accused Instrumentality." Patent L.R. 3-3(c) asks for the mirror image in invalidity contentions: a chart showing where in each item of prior art each limitation is found. Many other districts have similar rules.
Outside litigation, charts carry licensing discussions. A notice letter often arrives with charts for the strongest claims, and the recipient's counsel reads them looking for the row that fails.
The four parts of every row
Most charts use four columns, whatever they call them.
| Column | What it holds | What to check |
|---|---|---|
| Limitation | One element of the claim, quoted verbatim and labelled (1[pre], 1[a], 1[b]...) | The split is sensible and no words were dropped or paraphrased |
| Evidence | Quotations, figures, screenshots or test results, each with a locator | The source is about the right product, version and date |
| Finding | Present, partially present or not found | The label matches what the evidence actually shows |
| Reasoning | Why the evidence meets or misses the limitation | Every assumed meaning of a claim term is stated openly |
Limitation
The limitation column fixes what has to be proven. If two requirements share a row, a gap in one can hide behind evidence for the other. Check the labels against the claim and read each row word for word.
Evidence
Evidence should be specific enough that a reader could find it again: a document title, page and paragraph, a figure number, a timestamp in a video, a test log. A row that cites "product website" with no quotation is a placeholder, not evidence.
Finding
The finding is a label, and labels compress. "Present" can mean the evidence states the limitation outright, or that the author inferred it from two documents and an assumption. Read it last, after forming your own view.
Reasoning
The reasoning column is where the chart reveals its claim construction. Every finding depends on a reading of the claim terms, and construing patent claims is a matter of law for the court (Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996)). If the reasoning says that a "linked document" includes any database record that references another record, that is a construction, and the row stands or falls with it.
A worked example: claim 1 of US 6,285,999
US 6,285,999 B1, "Method for node ranking in a linked database", names Lawrence Page as inventor. It was filed on 9 January 1998, issued on 4 September 2001 and has expired. Below, its claim 1 is charted against a hypothetical intranet search appliance. The evidence quotations are invented for the example.
| Label | Limitation | Evidence (hypothetical) | Finding |
|---|---|---|---|
| 1[pre] | A computer implemented method of scoring a plurality of linked documents, comprising: | Admin guide §1: "The appliance ranks every page on your intranet." | Present |
| 1[a] | obtaining a plurality of documents, at least some of the documents being linked documents, at least some of the documents being linking documents, and at least some of the documents being both linked documents and linking documents, each of the linked documents being pointed to by a link in one or more of the linking documents; | Admin guide §3: "The crawler fetches each page and records every hyperlink between pages." | Present |
| 1[b] | assigning a score to each of the linked documents based on scores of the one or more linking documents and | Whitepaper p. 4: "Each page's rank score is computed from its inbound links, weighted by the importance of the page that links to it." | Partially present |
| 1[c] | processing the linked documents according to their scores. | Admin guide §5: "Search results are ordered by rank score." | Present |
The preamble row is charted even though whether a preamble limits a claim is itself a construction question. Charting it costs one row and avoids an argument about a missing one.
Rows 1[a] and 1[c] are straightforward, though a careful reader would still want a crawl log confirming that some pages both link and are linked to, because 1[a] requires it.
Row 1[b] is the one to study. The whitepaper shows that a page's score depends on the pages that link to it and on a property of each linking page. The limitation asks for more: the score must be based on the scores of the linking documents. "Importance" might be that score. It might also be something static, such as the linking page's depth in the site tree or a weight an administrator sets by hand. The evidence is close to the limitation without disclosing it, which is exactly what "partially present" should mean.
To close that row, the analyst needs evidence of how importance is calculated: source code, a technical paper, an engineer's testimony, or a test that changes one page's links and watches the scores of other pages move.
How to judge evidence quality
Ask the same questions of every row.
- Does the evidence state the limitation, or something near it? If the reasoning bridges a gap with paraphrase, the row is weaker than its label.
- Is it about the accused product? A different model, an older release or a supplier's reference design may not describe what the defendant sells.
- Is it from the right time? Infringement under § 271(a) must occur during the term of the patent. An invalidity reference must qualify as prior art, which depends on its date relative to the patent's filing or priority date.
- Is it technical material or marketing? A datasheet describes what a product does. A brochure describes what someone wants you to believe.
- How long is the chain of inference? One quotation that meets the limitation beats three documents joined by two assumptions.
- Are claim terms used consistently? If "linked documents" means web pages in row 1[a], it cannot mean database records in row 1[b].
Partial findings
A "partially present" row is not half a win. For literal infringement, a limitation that is not fully met is not met, so one partial row means the claim has not yet been shown to be infringed. The finding is useful because it points to the work that remains.
There are usually three ways forward: find better evidence, revisit the claim construction if the gap exists only under one reading of a term, or argue that the product contains an equivalent of the missing feature. Equivalence is a separate analysis with its own limits, and it deserves its own finding rather than a quiet upgrade to "present".
Dependent claims
Under 35 U.S.C. § 112(d), a dependent claim is construed to incorporate by reference all the limitations of the claim it refers to. A chart for a dependent claim 5 therefore needs every row of claim 1 plus the rows that claim 5 adds.
Two consequences follow. In an infringement chart, a dependent claim can be no stronger than its parent: if claim 1 is not met, claim 5 is not met either. In an invalidity chart the logic runs the other way. A dependent claim's added limitation can survive art that defeats the parent, which is why patent owners often assert dependent claims alongside the broad independent one.
Checks before you rely on a chart
- Count the rows against the claim. Every word of the claim belongs to some row.
- Read each row's evidence before its finding.
- List every construction the reasoning assumes, and mark the ones the other side will dispute.
- Mark every partial and not-found row. One is enough to defeat literal infringement of that claim.
- Confirm that each dependent claim carries all of its parent's rows.
Try it on your own patent
One Transform's Infringement search produces a first draft of this work. It starts from your patent's claims, parses them into individual limitations, and searches product documentation, datasheets and technical literature for products that appear to implement them. Each candidate is charted limitation by limitation, with a finding of present, partially present or not found and evidence for every element. Candidates are ranked by claim coverage, so a product that matches eight limitations out of ten outranks one that only sounds similar. Read those charts with the same checks you would apply to anyone else's.
Next step: run an Infringement search on one of your patents, or compare plans.
This article is general information about patent practice, not legal advice. For advice on a specific matter, consult a registered patent attorney or agent.
