US 6598791

I'll search for authoritative information on US 6,598,791, including any litigation/docket activity.

Let me run additional targeted searches for litigation and docket activity involving this patent.

Let me check litigation databases and the CAFC docket directly.

US Patent 6,598,791 — Analyst Summary

Data source note: The authoritative full text of this patent was provided in the working file (fetched from https://patents.google.com/patent/US6598791/en on 2026-09-30). I also ran live searches for USPTO records and litigation/CAFC dockets. Findings and gaps are noted at the end.

---

1. Bibliographic Data

Caveat: Google Patents itself states the listed assignee is not a legal conclusion. The assignment chain shows PSC Scanning → PSC Inc. → ECR Software Corporation, but "current assignee" should be confirmed in USPTO Assignment records before being relied upon.

---

2. Abstract (verbatim)

"A self-checkout system for a retail establishment that allows a customer to checkout multiple items having respective identification codes, the system includes a computer having memory with a buffer, an identification code reader coupled to the computer for determining the identifications of the items by the identification codes, a security verification mechanism coupled to the computer for verifying that the items actually being checked out from the retail establishment are the same as those identified by the identification code reader, and whereby the computer is adapted to store identification information of multiple items obtained by the identification code reader in the buffer before verifying that the items actually being checked out from the retail establishment are the same as those identified by the identification code reader."

---

3. Technical Problem and Core Disclosure

The patent addresses a throughput bottleneck in self-checkout: the prior art (FIG. 5A) required a scan → weigh → next item sequence, i.e., a security step interposed after every single scan. That step "is time-consuming, and can lead to … bottlenecks at the checkout stations, reduced throughput, unhappy customers, and negative impact on the financial condition of the establishment."

The disclosed solution is an item buffer 390 in main memory of the checkout computer 260 that temporarily holds identification information (e.g., SKU and retrieved weight from a Weight Look-Up database) for multiple items before any security verification occurs. Items in the buffer are described as "in hand" — the state between identification and placement on the security scale. Key disclosed capabilities:

• The number of items N scanned before verification may be fixed (e.g., two, three, four) or variable/adjustable by customer, attendant, manager, or external signal.
• Scanning and verification may be asynchronous so long as the buffer count is below N.
• Items may be verified in a different order than they were scanned; the system matches a weight differential against any item "in hand."
• Multiple items may be placed on the scale simultaneously and checked against the sum of the corresponding stored weights.
• On a match, the item(s) are cleared from the buffer and the count decremented; on a failure, an attendant is alerted and/or a re-weigh is performed.
• Alternatives to weighing are disclosed (camera-based image/dimension verification), as are EAS tag deactivation and a "reduced list" UI showing which items remain to be verified.

---

4. Independent Claims — Plain-Language Overview

The granted patent contains 40 claims, of which twelve are independent: 1, 10, 11, 14, 23, 26, 32, 35, 36, 37, 38, and 40. They fall into four families.

A. System claims

Claim 1 — generic checkout system (the broadest claim).
A checkout system for an establishment comprising: (a) a computer with memory that includes a buffer; (b) an identification code reader coupled to the computer for reading item codes; and (c) a security verification mechanism coupled to the computer that verifies the items actually being checked out are the same as those identified by the reader. The point of novelty is the functional limitation: the computer stores identification information for multiple items in the buffer before performing that verification.

Claim 10 — self-checkout variant, means-plus-function format.
The same idea for a retail self-checkout system, recited as a "means for reading," "security means," and "means for storing … in a buffer," i.e., §112(f)-style generic claiming.

Claim 11 — grocery-store-specific system.
A self-checkout system for a grocery store with a checkout terminal having memory with a buffer, a bar code scanner, and an electronic scale that verifies the weight of items actually being checked out matches the weight of items identified by the scanner. Again, the terminal stores identification information for multiple items in the buffer before scale verification.

B. Method claims

Claim 14 — generic checkout method. Reading identification codes of multiple items; storing identification information for the multiple items in a buffer; and verifying that one or more physical characteristics of the items actually checked out match the respective stored characteristic, after reading and storing codes for multiple items. (Broader than a weight-only claim because it recites any physical characteristic.)

Claim 23 — grocery self-checkout method. Reading bar codes of multiple grocery items; storing the identification information in a buffer; then verifying for security that weight of the items actually checked out matches the stored weights — the verification occurring after the multiple reads and stores.

Claim 26 — two-item sequence method. Reading and storing a first item's bar code, then — before verifying the first item's weight — reading a second item's bar code and storing it in the same buffer as the first item. The dependence on a shared single buffer is the distinctive limitation here.

Claim 32 — generalized two-item method. Reading the bar code of a first item, reading the bar code of a second item prior to any security verification of the first item, and then performing security verification on the first and second items. This is the plainest statement of the invention's core concept.

Claim 35 — out-of-order verification method. Consecutively reading bar codes of multiple items and storing identification information in a first order, then verifying the weights against stored information in a second order different from the first.

Claim 40 — combined-weight method (narrowest, most concrete). Scanning a first item's optical code; scanning a second item's code before placing the first on the scale; placing both items on the weigh scale; obtaining a combined weight; and using that combined weight to verify the identity of both items.

C. Computer program product claims

Claim 36 — a computer-readable medium with program logic adapted for the retail self-checkout system, comprising means for determining identification information for multiple items, means for storing that information in a buffer for multiple consecutive items, and means for verifying characteristics after multiple items are determined and stored.

Claim 37 — the same construct limited to a grocery store and to weight verification.

Claim 38 — a computer program product adapted to the two-item sequence of claim 26: reading/storing the first item, and reading/storing the second item into the same buffer before verifying the first item's weight.

---

5. Notable Drafting Irregularities (flagged, not corrected)

Consistent with the instruction to interpret identifiers literally:

• Claims 12 and 13 are each introduced as "A method according to claim 11…" / "A method according to claim 12…", yet claim 11 is a system (apparatus) claim. This is a claim-typing/dependency defect on the face of the granted patent. I have not auto-corrected it.
• Claim 12's body ("placing first and second items on a scale, determining an incremental weight increase … verifying that the incremental weight increase matches the sum of the weights") reads as a substantive independent-style method limitation appended to an apparatus claim.
• The pre-grant publication US 2002/0096564 lists a differently numbered claim set (e.g., its claim 12 is the granted claim 14, its claim 11 is the granted claim 11), indicating renumbering between publication and grant. Any citation to "claim 12" of this family must specify whether the publication or the granted patent is meant.

---

6. Prosecution / Prior-Art Context

The patent cites 52–57 references (the counts differ between the Google Patents "Citations (52)" list and its "Patent Citations (57)" list), including US 3,836,755 (Gretag), US 4,676,343 and US 4,792,018 (Checkrobot), US 4,940,116 (Checkrobot, unattended checkout), US 5,340,970 (article checkout with security parameter override), US 5,879,485 and US 6,092,725 (Symbol, statistical sampling security), and various NCR self-service terminal patents (US 5,747,784; US 5,952,642; US 5,966,264; US 6,036,128). The patent has a large forward-citation footprint (77 citing publications), including later NCR, Toshiba, Walmart, and PSC/ECR filings.

A closely related later patent is US 7,044,370 ("Checkout system with a flexible security verification system"), naming Donald C. Bellis and assigned to ECR Software Corporation, claiming priority to provisional 60/302,485 (filed 2001-07-02). Its specification incorporates by reference US 3,681,570 and 3,681,571 and US 5,340,970 regarding weight tolerances and manual override — useful context for the '791 family. Note this is a separate patent number, not to be conflated with 6,598,791.

---

7. Litigation and Docket Search — Results and Uncertainty

Finding: I did not locate any litigation or CAFC appeal involving US 6,598,791, and I located no CAFC 2026 docket referencing this patent.

• A search for "6598791" patent litigation CAFC 2026 docket returned zero results.
• Searches for "6,598,791" combined with infringement/courtlistener terms returned only unrelated cases (Neo Wireless, Realtime Data, ArrivalStar, Netlist/Samsung, Neonode '879 patent — a different patent whose last three digits coincidentally match).
• The only genuine appearance of US 6,598,791 in a litigation-adjacent document I found is as cited prior art, not as an asserted patent: it appears in the Declaration of Ivan Zatkovich filed as H-E-B Exhibit 1005 in IPR2020-00347 (Inter Partes Review of U.S. Pat. 9,934,506) and in related PTAB exhibits. That IPR is about a different patent (9,934,506); '791 is being used as prior art against it.

Explicit uncertainty — please treat as unverified:

1. I could not directly query USPTO PatentCenter or the CAFC docket (these are not reachable through the search tool available here). My litigation conclusion rests on indirect search results only.
2. Absence of results is not proof that no suit or appeal exists. A 2001 patent that expired 2021-01-19 could still be the subject of a live CAFC appeal in 2026 (e.g., an appeal from a district court or PTAB decision on pre-expiration conduct, or from an IPR). I found no such matter, but I cannot rule one out.
3. Assignee/ownership statements are drawn from Google Patents' graded "assumption" data and the recorded assignment events; they should be confirmed against USPTO Assignment records.

---

8. Bottom Line

US 6,598,791 is a PSC Scanning-originated, later ECR Software-assigned patent (Bellis/Keys/Behrens, filed and priority 2001-01-19, granted 2003-07-29, expired 2021-01-19) that claims the now-ubiquitous self-checkout architecture in which multiple items are scanned into an item buffer before any weight/security verification, with verification permitted to occur asynchronously, out of scan order, or on combined item weights. Its twelve independent claims span systems (1, 10, 11), methods (14, 23, 26, 32, 35, 40), and computer program products (36, 37, 38). The specification is remarkably predictive of modern self-checkout practice, but the patent is expired and I found no evidence of any pending litigation or CAFC appeal as of this date.

Primary citations: https://patents.google.com/patent/US6598791/en (full text, claims, assignments, citations); https://www.freepatentsonline.com/6598791.html (claim text and cited references); https://www.freepatentsonline.com/[7044370](/patent/7044370).html (related Bellis/ECR patent); PTAB IPR2020-00347, Exhibit 1005 (showing '791 cited as prior art).

I'll search for litigation involving US patent 6598791.

The initial results are pulling in a different patent — Neonode's '879 touch patent (a smartphone patent), not US 6,598,791. Let me search more precisely.

The search results keep conflating this with Neonode's touch patent ('879) and Walmart's mobile self-checkout patents — neither is US 6,598,791. Let me search litigation databases directly.

Bottom line

I found no litigation — no district court case, no CAFC appeal, and no PTAB proceeding — in which US Patent 6,598,791 (the "Self-checkout system and method including item buffer for item security verification" patent) was asserted as a patent-in-suit.

I want to be explicit about that, because a naive keyword search will hand you a lot of false positives (see the warning below). I could not locate a single case matching the required fields (plaintiff / defendant / jurisdiction / case number / filing date / outcome).

Why the obvious searches return garbage

Searching "6598791" or "the '879 patent" produces a flood of hits that are not this patent:

Per your rule, I am not auto-correcting any of those numbers — I am simply flagging that they must not be mistaken for US 6,598,791.

The one "hit" that involves this exact patent (and why it isn't litigation)

US 6,598,791 does appear in a PTAB-adjacent document — but as prior art, not as the challenged or asserted patent:

• IPR2020-00149, Inter Partes Review of U.S. Pat. No. 9,262,781 — the patent owner's exhibit (Exhibit 1005, Zatkovich Declaration, Appendix C) reproduces the full text and abstract of US 6,598,791 as a prior-art reference.

That is an inter partes review of a different patent in which the '791 patent was cited. It is not a validity challenge to US 6,598,791, and the '791 patent was not a party to it.

Prosecution citation trail (also not litigation)

Google Patents' "Cited By" list for US 6,598,791 is long (e.g., NCR, Walmart, Toshiba, Evolution Robotics self-checkout patents), but every one of those is a prosecution citation — later applicants citing '791 as background art. None reflects a lawsuit.

What I can confirm about the asset itself (from the authoritative patent record)

• Patent: US 6,598,791 B2; App. No. 09/766,392; filed 2001-01-19; granted 2003-07-29; priority 2001-01-19.
• Original assignee: PSC Scanning, Inc. → PSC Inc. / ECR Software Corp. (2004 assignment). Current assignee of record listed as PSC Inc USA and ECR Software Corp.
• Family members do not include litigation: EP 1356425 A1 (status: withdrawn), WO 2002058003 A1 (PCT/US2002/004555, status: ceased).
• Legal status: Expired – Lifetime, with anticipated expiration 2021-01-19.

Caveats / confidence

• I searched general web indexes, CourtListener-hosted RECAP documents, the Unified Patents portal, and litigation-complaint aggregators. None returned a case asserting US 6,598,791.
• I did not obtain a direct PACER docket query or a targeted Unified Patents "litigation/caselist" pull filtered on this exact number (I was cut off before completing that step). It is therefore possible that a short-lived or recently filed case exists that is not well-indexed. If you need certainty for an enforceability/freedom-to-operate opinion, a PACER docket search and a Docket Alarm / Lex Machina search on the literal string "6,598,791" is warranted.
• Given the patent expired January 19, 2021, the practical likelihood of new assertions is essentially nil; any infringement activity would have to predate expiry.

Conclusion: On the available record, no litigation involving US 6,598,791 is known. Any case you see captioned with "the '879 patent" involving Neonode, Apple, or Samsung is a different patent and must not be attributed to this one.

I'll verify the structured block's "no proceedings" default against public sources before writing this up.

No PTAB hits yet. Let me probe a few more angles — litigation/assertion history and defensive aggregators.

Let me check assertion/litigation history and try one more targeted PTAB query format.

Proceedings overview

Total AIA trial proceedings on file for US 6,598,791: 0 — the structured "PTAB proceedings on file" block returns no IPR, PGR, or CBM for this patent from the USPTO Open Data Portal, and my web searches (Google Patents record, PTAB decision/paper pages, aggregator results) surfaced no proceeding number, no petition, no institution decision, and no Final Written Decision naming the '791 patent. Breakdown by status is therefore: active 0 / claims invalidated 0 / claims sustained 0 / settled 0 / institution denied 0.

Bottom line for a defendant: this is not a "hardened by surviving IPR" story and not a "claims canceled" story — it is a no-challenge-history story, because the patent expired 2021-01-19 (Google Patents legal status: "Expired – Lifetime"; anticipated expiration 2021-01-19) before any challenger had an economic reason to file. All 40 claims are untested at the PTAB. The defensively useful fact is not the IPR record — it is the expiry date and the § 286 damages window that follows from it. No proceeding is available to itemize below.

Verification caveat (stated plainly): I could not run a live PTAB E2E / PTAB Decisions docket query or a Docket Alarm / CourtListener docket pull — my tool calls were exhausted after the searches above. The ODP block is the canonical source and it is empty; my independent searches are consistent with that. If you need a belt-and-suspenders confirmation before filing, query the patent number directly in PTAB E2E (https://e2e.uspto.gov) and PTAB Decisions (https://developer.uspto.gov/ptab-web/#/search/decisions).

No proceedings to itemize

There is no {PROCEEDING_NUMBER} — {Petitioner} v. {Patent Owner} entry to write, because no petitioner ever filed against this patent. Accordingly I am not supplying judge panels, petition grounds, institution reasoning, FWD claim-level dispositions, settlement terms, or CAFC docket numbers — none of those exist, and inventing them would be fabrication.

---

Strategic summary

Claim status — CANCELED vs. SUSTAINED vs. UNTESTED. The patent carries 40 claims (issued as US 6,598,791 B2 on 2003-07-29 from App. No. 09/766,392, filed 2001-01-19, claiming priority to 2001-01-19; inventors Donald Bellis, Jr., Barry Keys, and James Behrens; original assignee PSC Scanning, Inc., later ECR Software Corporation). Canceled: none. Sustained after challenge: none. Untested: all claims 1–40. The independent claims are the apparatus/system claims 1, 10, and 11; the method claims 14, 23, 26, 32, 35, and 40; and the computer-program-product claims 36, 37, and 38. The commercially interesting ones for a self-checkout defendant are claim 1 (buffer-before-verification system), claim 14 (read/stores multiple, then verify physical characteristic), claim 26 (store second item in the same buffer before verifying the first), claim 32 (read second item prior to verifying first), and claim 40 (combined-weight verification of two scanned items). Every one of them is unadjudicated. There is no FWD to link to and nothing to quote.

Estoppel landscape. Because no IPR was ever instituted and no FWD ever issued, § 315(e)(2) estoppel is entirely absent. No petitioner (and no privy) is barred from raising any § 102/§ 103 ground, and there is no IPR-estoppel overlay constraining your invalidity case in litigation. Separately, the absence of any petition means there is also no § 315(b) one-year bar running against you, and no § 325(d) "same or substantially the same art" discretionary-denial problem created by an earlier Board record. A defendant's prior-art position is completely unencumbered — subject only to the practical reality that the patent is expired, which makes the IPR route moot (see below).

Pattern signals. No repeat petitioner exists (there is no first petitioner). No defensive aggregator — Unified Patents is not in this chain on any record I could find; the Unified activity my searches surfaced was on unrelated patents (e.g., claims 1–7 of the '228 patent, Dynapass '658, and the '158 patent), none of them the '791 patent. The patent owner has not pursued any PTAB appeal, because there is no PTAB proceeding to appeal. The only post-grant-adjacent events of record are foreign prosecution outcomes on the family — EP 1356425 A1 (nationalized from EP 02718988A) went abandoned/"withdrawn" and WO 2002058003 A1 "ceased" — plus two US assignment recordings (2003-12-29 and 2004-05-28) between PSC Scanning, PSC Inc., and ECR Software Corporation. Note the Google Patents "current assignee" field lists both PSC Inc USA and ECR Software Corp; treat the chain-of-title as needing a real assignment-record pull before you name a defendant-side counterparty. Also note a related but distinct patent, US 7,044,370 (Bellis, "Checkout system with a flexible security verification system," provisional priority 60/302,485, filed 2001-07-02), sits adjacent to this family — do not conflate it with the '791 patent, and do not assume an IPR on the '370 patent is an IPR on the '791.

One structural vulnerability worth flagging if you are ever asserted against. The claim set has drafting errors: claim 12 recites "A method according to claim 11," claims 24–25 recite "A method of claim 23" / "A method according to claim 24," and claim 34 recites "A method of claim 32" — yet claim 11 is an apparatus claim and claim 32 is a method claim. Claims 12–13 thus depend from an apparatus claim while purporting to be method claims. These are § 112 ¶ 2 / dependency-definiteness arguments, not invalidity certainties, but they are free ammunition that no tribunal has yet considered.

---

Recommended next steps

1. Lead with expiry, not with the IPR record. The patent's term ended 2021-01-19. For any conduct after that date there is no infringement exposure, and a demand letter that accuses post-2021-01-19 self-checkout systems should be answered with the face of the patent's legal-status field: "Expired – Lifetime," anticipated expiration 2021-01-19. There is no FWD to link to because none exists — the disposition you should quote is the expiration date, from the Google Patents record (https://patents.google.com/patent/US6598791/en).

2. Run the § 286 math before you spend anything. Under 35 U.S.C. § 286 no damages may be recovered for infringement more than six years before the complaint is filed. A hypothetical complaint filed 2026-09-30 would reach back only to 2020-09-30; combined with the 2021-01-19 expiry, the entire recoverable window is roughly 2020-09-30 through 2021-01-19 — under four months, and only for a defendant who was actually selling an accused system then. Push the filer to identify pre-expiry accused conduct with specificity; a demand built on current product line-ups is value-less.

3. Because the patent is expired, an IPR is almost certainly not worth filing. The Board can institute on expired patents (the patent remains eligible for review, though there is no prospective exclusionary relief to win), but with no live damages window and no pending assertion, a § 311 petition buys you estoppel exposure and fees for a decision that cannot enjoin anyone. If a suit is later filed on pre-expiry conduct, re-evaluate: at that point an IPR is still available (no § 315(b) bar has been triggered by any earlier petition) and no § 315(e) estoppel exists to limit it.

4. Preserve the full prior-art record now. Since no proceeding has ever been filed, nothing is off the table — no § 325(d) baggage, no estoppel, no IPR-art narrowing. The references cited on the '791 face (Checkrobot, NCR, Symbol, Schneider, Humble, Dumont, and the Teraoka/BIT self-scanning checkout art) are all still available as § 102/§ 103 art, and there is a large third-party citation set (77 citing documents) that reflects the field's dense prior art.

5. If you still want a docket-level confirmation, do these two lookups. (a) PTAB E2E — https://e2e.uspto.gov, search party/patent "6,598,791"; (b) PTAB Decisions search — https://developer.uspto.gov/ptab-web/#/search/decisions. For any district-court assertion history, run CourtListener RECAP — https://www.courtlistener.com for "ECR Software" as plaintiff. If those return empty, the record stands as reported here: no PTAB activity on file, all 40 claims untested, patent expired 2021-01-19.

I'll search for the actual assignment records before writing anything, since reel/frame numbers must not be fabricated.

Found a concrete reel/frame lead. Let me chase the remaining links.

Assignment / Ownership Reconstruction — US 6,598,791 B2

Data-source caveat (read first). My tool budget for this task was exhausted after four search rounds, so I could not run a live query against USPTO Assignment Center (https://assignmentcenter.uspto.gov/ or https://assignment.uspto.gov/patent/index.html). What follows is built from (a) the authoritative Google Patents legal-events block for US 6,598,791 supplied in the working file, and (b) one primary USPTO assignment document image I did retrieve (reel 014822/0408, the PSC Inc. IP security agreement), plus corroborating records. Where I do not have a reel/frame, I say so rather than supplying a number. Three of the five recorded events below therefore have no reel/frame confirmed — that is a gap, not a finding of absence.

---

Inventors

Pattern observations (and non-observations):

• No "all inventors departed within 12 months" pattern. The opposite is documented: Bellis and Keys were still filing as PSC Scanning, Inc. inventors in July 2001 — WO 03/005313 A3 ("Checkout system with a flexible security verification system," Bellis/Keys/Denimarck) and WO 03/009600 A1 ("Video identification verification system," Bellis/Denimarck/Keys), both PSC Scanning applicants. That is contemporaneous with, and shortly after, the '791 filing. So the two-inventor core stayed with the assignee through at least mid-2001.
• Name-order discrepancy between publication and grant. The pre-grant publication US 2002/0096564 A1 is indexed as "Behrens et al.", while the granted patent lists Bellis first. This is a § 1.76/1.63 sequencing artifact, not an ownership event — but it matters if you are matching the publication to the grant in a chain-of-title search.
• Behrens's later trajectory is unverified. He does not appear on the later PSC self-checkout applications I surfaced (WO 03/005313, WO 03/009600). I have no evidence he left PSC, and no departure date — flagging this as unclear, not as a pattern.
• Nothing in the record indicates a founder-inventor consultancy or an inventor-held residual. All three assigned to PSC Scanning at filing.

---

Original assignee

PSC Scanning, Inc. — an operating company, and a real one: a manufacturer of fixed and handheld bar-code scanning hardware (the '791 specification itself touts the Magellan® scanner "available from PSC Inc. of Eugene, Oreg." — the applicant citing its own product).

• Product embodying the claims: Yes, at least in hardware form. The specification describes the claimed system as built around PSC's own multi-scan window scanner, i.e., the patent covers a self-checkout station architecture productized by the assignee's scanner line.
• Primary line of business: Automatic identification and data capture (AIDC) — bar-code scanners, readers, verifiers. PSC's trademark schedule in the security agreement confirms the goods/services: "bar code readers," "bar code scanners," "optical and electronic instruments."
• Current status: Not established. The corporate form changed from PSC Scanning, Inc. to PSC Inc. (the 23 Dec 2003 security agreement identifies the grantor as "PSC Inc., a New York corporation"), and the patent was then assigned out to ECR Software Corporation on 2004-05-28. I did not verify whether PSC Inc. was later acquired, dissolved, or reorganised — do not rely on any acquisition story I have not sourced.
• Current assignee of record ambiguity: Google Patents' "Current Assignee" field lists both "PSC Inc USA" and "ECR Software Corp." That dual listing is itself a signal that the terminal title position is not cleanly resolved in that dataset, and it should be resolved against Assignment Center's "current owner" view before anyone is named in a demand or a license.

---

Assignment timeline

Five recorded events bear on this patent. Reel/frame is confirmed for two; for the other three I state "not surfaced" rather than guess.

---

2001-01-19 (executed, concurrent with filing) / recorded 2001-01-19 — Reel 011496/0319

• Conveyance: Assignment of assignor's interest (original)
• Assignor: Bellis, Don; Behrens, James; Keys, Barry (as originally recorded — see the 2001-06-01 correction)
• Assignee: PSC Scanning, Inc.
• Correspondent: not surfaced. (Patent's attorney of record as printed on the face is MacCord Mason PLLC, Greensboro, NC — but I did not confirm MacCord Mason was the recording correspondent on this reel.)
• Context: Inventor-to-employer assignment at filing — normal operating-company capture of employee invention rights.

Sourcing note: the reel/frame 011496/0319 is confirmed indirectly but reliably — Google Patents' own 2001-06-01 entry states the corrective assignment corrected the assignor name "PREVIOUSLY RECORDED AT REEL 011496, FRAME 0319." I did not view the underlying image.

---

2001-06-01 (executed) / recorded 2001-06-01 — Reel not surfaced

• Conveyance: Correction (corrective assignment to correct the assignor name)
• Assignor: Bellis, Jr., Donald; Behrens, James; Keys, Barry
• Assignee: PSC Scanning, Inc.
• Correspondent: not surfaced
• Context: Scrivener's-error correction only — "Don Bellis" → "Donald Bellis, Jr." No change in ownership, no change in assignee. Not an NPE signal.

---

2003-12-23 (executed) / recorded 2003-12-31 per the USPTO receipt stamp (Google Patents event date: 2003-12-29) — Reel 014822/0408 (agreement body at frames 0421–0424)

• Conveyance: Security Agreement — "Intellectual Property Security Agreement" in favour of Agent for the Lenders. Not a transfer of title.
• Assignor / grantor: PSC, Inc., a New York corporation
• Assignee / secured party: Wells Fargo Foothill, Inc., as Agent (Atlanta, GA 30328)
• Correspondent: Penelope J. A. Agodoa, Federal Research Company, LLC, 1030 15th Street NW, Suite 920, Washington, DC 20005; PTO-1595 cover sheet signed Michael A. Witt, Esq.
  • Recurrence flag: this is the only correspondent I recovered anywhere in this chain, and it appears once — on a third-party IP recording-service filing, not on any title transfer. A single appearance is not the signal; the signal would be recurrence, and I found no recurrence in this chain. Federal Research Company is a docketing/recording service, not an entity I associate with NPE assertion.
• Context: Securitization — collateral pledge. The cover sheet recites 80 patents total in Exhibit A, secured against a loan facility (jury-trial waiver; New York venue; "Agent's Right to Sue" upon default). This is a lender's blanket IP lien, characteristic of a leveraged operating company.

Two verification caveats on this entry: (i) Exhibit A was not directly viewed — the presence of '791 among the 80 pledged patents is inferred from the Google Patents legal-event entry, not confirmed from the exhibit list. (ii) No release of this security interest was found for '791. A release-by-secured-party chain for other PSC-family IP appears in a TTAB record (reels 3449/0742 recorded 2006-12-22, 3944/0808 recorded 2009-03-02, 3957/0417 recorded 2009-03-23, 6564/0693 recorded 2019-02-14, naming Wells Fargo Capital Finance LLC fka Wells Fargo Foothill as releaseor) — but I could not tie any of those reels to the '791 patent, so treat the release question as OPEN. It is genuinely possible that a technically-live security interest sits on the record.

---

2003-12-29 — Reel not surfaced

• Conveyance: Assignment of assignor's interest (as recorded in the dataset)

• Assignor: ECR Software Corporation

• Assignee: PSC Inc.

• Correspondent: not surfaced

• Context: FLAGGED — apparent record anomaly; stating it literally, not correcting it. Two independent inconsistencies:

  1. The direction (ECR Software → PSC Inc.) is the reverse of the 2004-05-28 record (PSC Scanning → ECR Software), producing an implausible round-trip.
  2. The security agreement executed six days earlier (2003-12-23) already recites PSC, Inc. as the fee owner/grantor, which is inconsistent with PSC Inc. acquiring the patent on 2003-12-29 from ECR Software.

  A second, more coherent reading of the same event is that it reflects an internal reorganisation / change of name by which PSC Scanning, Inc. → PSC, Inc. took effect on that date, with the dataset swapping the assignor/assignee fields. I cannot adjudicate between the two without the underlying document. Either way, no external third party entered the chain on this date.

---

2004-05-28 (executed) / recorded 2004-05-28 — Reel not surfaced

• Conveyance: Assignment of assignor's interest
• Assignor: PSC Scanning, Inc.
• Assignee: ECR Software Corporation
• Correspondent: not surfaced
• Context: Divestiture — the terminal recorded title transfer. Note the assignor is named as PSC Scanning, Inc., not PSC Inc., which is a second reason to treat the 2003-12-29 entry above as a data artifact rather than a real ECR→PSC conveyance. This is the last title event of record; every later dataset entry (including the dual "PSC Inc USA / ECR Software Corp" current-assignee listing) is downstream of it.

---

2021-01-19 — not an assignment; anticipated expiration (20 years from filing). Recorded here because it terminates the commercial relevance of the ownership chain.

---

Timeline diagram

timeline
    title Ownership of US 6598791
    2001 : Filed by PSC Scanning Inc
         : Inventors assign rights to PSC Scanning
         : Corrective inventor name assignment recorded
    2003 : Patent issues as US 6598791 B2
         : Security agreement to Wells Fargo Foothill
         : Possible internal PSC name change
    2004 : Assignment to ECR Software Corporation
    2021 : Patent term expires

---

NPE / troll-pattern signals

---

Verdict

Insufficient data.

On the record I can actually source, the chain runs employee inventors → PSC Scanning, Inc. (2001-01-19, reel 011496/0319) → [correction 2001-06-01] → PSC, Inc., which pledged 80 patents to Wells Fargo Foothill as collateral (executed 2003-12-23, reel 014822/0408) → ECR Software Corporation (2004-05-28). That is an operating-company-to-operating-company chain with a lender's lien in the middle — zero of the eight NPE signals is affirmatively met, and the one clustering that looked interesting (three recordings in five months) is explained by a name change plus a secured financing rather than LLC layering. I decline to call this "NPE — moderate confidence" because the constraint that governs here is that "could be a shell" is not a finding, and nothing in the record makes ECR Software Corporation look like one. Equally, I cannot call it "Operating-company assertion" — that verdict requires a current assignee shipping products and suing competitors, and there is no suit. And I cannot call it "Defensive / non-asserting," since no defensive aggregator appears. The honest verdict is therefore insufficient data, driven by three concrete gaps: (i) reel/frame not surfaced for the 2001-06-01 correction, the ambiguous 2003-12-29 entry, and the 2004-05-28 sale; (ii) Exhibit A of reel 014822/0408 not read, so the pledge of '791 itself is inferred; (iii) no release of the Wells Fargo security interest located, so the encumbrance status of the final link is unknown.

Verification links. Assignment Center search: https://assignmentcenter.uspto.gov/ · legacy Assignment Search: https://assignment.uspto.gov/patent/index.html (query patent number 6,598,791). The primary security-agreement image I relied on is https://legacy-assignments.uspto.gov/assignments/assignment-pat-014822-0408.pdf.

Open items to close before anyone relies on this chain — (1) pull the reel/frame and correspondent for the 2004-05-28 transfer and confirm the assignee's exact legal name; (2) read Exhibit A of reel 014822/0408 to confirm '791 was among the 80 pledged patents; (3) search for a release by secured party from Wells Fargo Foothill / Wells Fargo Capital Finance covering this patent family, and if none exists, treat the lien as standing; (4) resolve the 2003-12-29 entry against the underlying document to determine whether it is a title transfer or a PSC name change; (5) confirm whether ECR Software Corporation still exists and what it does — that single fact is what separates "operating successor" from "orphan asset," and I could not establish it here.

I'll verify the patent record and then examine the cited prior art references in detail.

Let me gather details on the remaining key self-checkout references.

I've hit the tool-call ceiling, so I'll write up the prior-art analysis now, drawing on (a) the authoritative full text in the working file and (b) the live search results I did obtain before the limit.

---

Prior-Art Analysis — US 6,598,791 B2

0. Record confirmation first

The number 6,598,791 resolves to exactly one patent and nothing else:

Because the application has a § 102 effective date of 2001-01-19, the governing law is pre-AIA 35 U.S.C. § 102 (a), (b), and (e). All analysis below uses that framework. Nothing in the search results contradicted the working-file text, so I treated the working file as authoritative and used the searches only to confirm reference content (Checkrobot, Symbol, NCR/Checkrobot quoting, etc.).

Anticipation standard applied: a reference anticipates a claim under § 102 only if a single reference discloses every limitation, arranged as in the claim. Where a reference discloses most but not all, I say so and route it to § 103 rather than calling it anticipatory.

---

1. The threshold observation that governs everything

The patent's own specification concedes, and illustrates at FIG. 5A, that the "scan → weigh → repeat" per-item self-checkout loop was old — "each item to be purchased was sequentially scanned at step 500, and weight verified at step 510 before the next item could be processed." That means the entire body of cited Checkrobot/Schneider/NCR/Tec self-checkout art is aimed at that per-item loop, and therefore most of the 57 references do not reach the one limitation that carries the granted claims: storing identification information for multiple items in a buffer before any security verification.

So the honest tiering is:

• Tier 1 (the only real § 102 candidates): references that arguably put more than one item's identity at rest before a characteristic check.
• Tier 2 (§ 103 fodder / element-level support): the scale + scanner + stored-weight self-checkout family.
• Tier 3 (background only): commerce/security/portable-scanner art that supplies no buffer-before-verify teaching.

---

2. Tier 1 — Closest § 102 candidates

2.1 US 4,792,018 — Humble et al. / Checkrobot Inc.

• Citation: US 4,792,018 A, "System for security processing of retailed articles."
• Dates: priority/filed 1984-07-09 (CIP of Ser. No. 628,913); published 1988-12-20. (Confirmed live.)
• Disclosure: Customer scans a UPC; article is placed on an entry conveyor running through a "security tunnel" bounded by light curtains; the weight (and/or size) of the article is measured in the security zone and compared to a stored characteristic fetched using the scanned UPC; the conveyor reverses if comparison fails. Critically, the specification states the system "can also accommodate plural articles in residence on the entry conveyor and effecting the weighing scale at common times," and the controller "stores a signal indicative of a predetermined value of the measurable article characteristic correlated with such article identification code."
• § 102 mapping: This is the single strongest cited reference against the "multiple items" + "weighed in combination" concepts.
  • Potentially anticipates claim 40 (scan item 1; scan item 2 before weighing item 1; place both on the scale; obtain a combined weight; use it to verify both) via the express "plural articles… weighing scale at common times" passage.
  • Potentially anticipates claim 1 / claim 14 only if one maps the controller's correlation store (UPC ↔ characteristic, plural articles) onto the claimed "buffer" and treats the tunnel measuring step as the "security verification mechanism" acting after plural items are resident. That is a genuinely arguable but not clean reading — the plain embodiment verifies each article as it transits.
  • Does not anticipate claim 35 (out-of-order verification), because verification is conveyor-ordered.

2.2 US 4,676,343 — Humble, Gentzler et al. / Checkrobot Inc.

• Citation: US 4,676,343 A, "Self-service distribution system." (Family member GB 2,193,000 A; published 1987-06-30, priority 1984-07-09.)
• Disclosure: Customer scans UPCs of presented articles, articles are conveyed out of reach through a light-curtain tunnel, a physical parameter is cross-checked against the scanned identification data, failures are returned. On completion of scanning the customer "activates an input signal and is furnished with a printed itemized list," which is then taken to a cashier.
• § 102 mapping: The compilation of a multi-item scanned list before the later cashier/payment step is a plausible § 102 hook for claims 14 and 32 ("read second item prior to performing security verification that the first item is the same"). The cross-check, however, is performed article-by-article on the conveyor, so claims requiring verification after plural reads with the item held "in buffer" are supported only obliquely.

2.3 US 5,877,485 and US 6,092,725 — Swartz et al. / Symbol Technologies

• Citations: US 5,877,485 A, "Statistical sampling security methodology for self-scanning checkout system" (published 1999-03-02, priority 1996-01-25); US 6,092,725 A, same title (published 2000-07-25, priority 1997-01-24). (Content confirmed live — Swartz/Shellhammer/Katz et al., Symbol Technologies.)
• Disclosure: A portable self-scanning terminal compiles a list of all items scanned by the shopper while shopping; at checkout a cashier/audit step re-scans a statistically selected subset n of items and compares them to the compiled list; the transaction is allowed only if each re-scanned item is present on the list and disallowed otherwise.
• § 102 mapping: This is the most conceptually uncomfortable reference, because it is a genuine scan-many-into-a-buffer-then-verify-later architecture.
  • Potentially anticipates claim 1 / claim 10 if the shopper's compiled list is treated as the "buffer," the scanner as the "identification code reader," and the audit re-scan as the "security verification mechanism … verifying that the items actually being checked out … are the same as those identified." That is a real § 102 argument the '791 examiner evidently did not run.
  • Does not anticipate claims 14, 19, 23, 26, 27, 30, 40, because those require verification of a physical characteristic/weight matching a stored characteristic associated with the identification information — Symbol verifies by re-reading the code, not by weighing.
  • Does not anticipate claim 35 (different verification order) — the audit is not order-scrambled against the list.

2.4 US 4,940,116 — Checkrobot Inc.

• Citation: US 4,940,116 A, "Unattended checkout system and method." Published 1990-07-10, priority 1989-03-07.
• Disclosure: Unattended conveyorized checkout with security processing; article-by-article.
• § 102 mapping: Element-level support only (code reader + stored characteristic + rejection). Does not teach a multi-item identity buffer before verification. § 103 material against claim 1/11 at best.

2.5 US 5,340,970 — Wolfe, Allard, Nguyen et al. / Checkrobot Inc.

• Citation: US 5,340,970 A, "Article checkout system with security parameter override capacity." Published 1994-08-23, filed/priority 1992-03-17; assignee CheckRobot Inc. (Content confirmed live, including the "override capacity" claim set.)
• Disclosure: UPC reader + sensor (weight and/or size) + memory storing a standard characteristic correlated with UPC + comparator + rejector; operator override substitutes the freshly measured characteristic into the store.
• § 102 mapping: The override disclosure is the natural § 103/§ 112 companion to the '791 re-weigh/attendant-override passages, but it is a single-item verification loop. It does not anticipate any independent claim of '791.

---

3. Tier 2 — Weight/scale self-checkout family (§ 103 fodder, element support)

These collectively establish every element except the multi-item buffer, i.e., they are the FIG. 5A art:

§ 102 conclusion for Tier 2: none of these anticipates an independent claim of '791, because each verifies one item at a time and none recites holding identities for multiple items in a buffer before verification. They are, however, the natural § 103 combination base if one pairs any of them with the Tier-1 multi-item disclosures.

---

4. Tier 3 — Background / non-analogous cited art (no § 102 force)

---

5. Reference-date anomalies the analyst must flag (not auto-corrected)

1. JP 2001-184563 A published 2001-07-06, i.e., after the '791 critical date of 2001-01-19. As a foreign publication it is therefore not § 102(a)/(b) art against '791 on its face. It can only be relevant if a US counterpart has an earlier effective filing date giving § 102(e) status. This warrants verification against the JP/US family.
2. US 6,354,498 and US 6,363,355 granted in 2002, i.e., after the '791 filing. They are citable only under § 102(e) (pre-AIA), via their 1997/1998 US filing dates — not as printed publications.
3. The Google Patents record shows the cited-reference count as both "Citations (52)" and "Patent Citations (57)," and separately lists Stambler-family members (5,524,073 / 5,646,998 / 5,793,302 / 5,974,148 / 5,936,541) that appear in only one of the two lists. Treat 52 vs. 57 as an indexing discrepancy, not a legal difference; the union of both lists is reproduced above. I did not auto-reconcile the two numbers.

---

6. Claim-by-claim § 102 anticipation map (best available case)

---

7. Bottom line on the '791 prior art

• Most relevant cited references, ranked: (1) US 4,792,018 (Checkrobot) for plural articles weighed at common times; (2) US 4,676,343 (Checkrobot) for scan-many-then-list; (3) US 5,877,485 / US 6,092,725 (Symbol) for scan-many-into-a-list-then-verify-later; (4) US 5,340,970 (Checkrobot) for stored-characteristic + override; (5) the NCR weight-scale self-checkout family as the express FIG. 5A art.
• No single cited reference anticipates the full independent-claim set, and none discloses out-of-order verification (claim 35) or express "same buffer" sequencing (claim 26). The granted claims likely survived because the examiner treated the cited art as the per-item loop the patent criticizes in FIG. 5A.
• The strongest § 102 attack available from this face-of-the-patent art is against claim 40 (via '018's plural-article/combined weighing) and claims 1/32 (via Symbol's scan-many-then-audit). Everything else is a § 103 combination case.

---

8. Verification caveats (stated plainly, per operating rules)

1. I exhausted my tool calls before independently reading the full texts of several Tier-2 references (e.g., US 5,952,642; US 5,996,264; US 6,032,128; US 4,940,116; US 5,609,223; US 5,992,570). For those I relied on titles, assignees, dates, and short excerpts surfaced in the searches; the "brief descriptions" above are therefore title-and-context-derived, not full-text-verified. Treat them as leads to be confirmed against the PDFs.
2. The § 102 mapping for claims 1/10/14/32/40 rests on claim-construction calls (is a re-scan audit a "security verification"? is a conveyor a "buffer"?) that are genuinely contestable. I have labeled them "arguable," not "anticipatory," to avoid overclaiming.
3. The Google Patents citation lists (52 vs. 57) are not reconciled, and the JP 2001-184563 / US 6,354,498 / US 6,363,355 post-filing dates require family verification before any of them is cited as § 102 art.
4. Nothing I found contradicts the working-file text; where the search results and the file overlapped (the '018, '343, '485, '970 texts), the search results confirmed the file.

Primary sources: https://patents.google.com/patent/US6598791/en (full text, claim set, citation lists); https://patents.google.com/patent/US4792018 (Checkrobot, plural-article weighing); https://patents.google.com/patent/[US4676343A](/patent/US4676343A) (Checkrobot, self-service distribution); https://www.everypatent.com/comp/pat5340970.html (Checkrobot '970 override); patentimages/US20010015375A1.pdf (Swartz et al., Symbol statistical-sampling self-checkout).

I'll ground this analysis by pulling details on the key prior-art references that appear on the face of US 6,598,791 before writing the § 103 analysis.

Excellent — the Schneider '465 reference is highly probative. Let me pull a few more key references to complete the picture.

Obviousness Analysis of US 6,598,791 under 35 U.S.C. § 103

Analyst note on scope and evidence. This section builds on the patent summary, litigation summary, PTAB summary, and strategic summary already produced; I do not repeat their content. I use the prior-art references listed on the face of US 6,598,791 (its "Citations"/"Patent Citations" lists, plus the examiner-cited art reflected in the Google Patents record at https://patents.google.com/patent/US6598791/en). I ran live searches to verify the actual disclosure of the four most probative references. Where I am working from a title only, I say so expressly rather than attributing disclosure to it.

Flagged inconsistency. The current task header states "Current Date: April 26, 2026," while the earlier-generated sections state the working file was fetched 2026-09-30 and computed the § 286 window from that date. I have not auto-corrected either; treat the § 286 arithmetic in the earlier section as keyed to 2026-09-30.

---

1. Governing legal standard

• Graham v. John Deere Co., 383 U.S. 1 (1966): scope and content of the prior art; differences between the prior art and the claims; level of ordinary skill; objective indicia.
• KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398, 416–17 (2007): "the combination of familiar elements according to known methods is likely to be obvious when it does no more than yield predictable results"; where a technique has been used to improve one device, a skilled artisan who recognizes it will improve similar devices in the same way; a finite number of identified, predictable solutions amounts to "obvious to try."
• MPEP §§ 2141–2143 (rationales: combining prior art elements per known methods, substitution of known elements, use of known technique to improve similar devices, obvious to try, design incentives/market forces).

Because the patent expired 2021-01-19 and was never litigated or challenged (per the earlier sections), there is no objective-indicia record and no prosecution-estoppel overlay constraining this analysis.

---

2. Level of ordinary skill in the art (POSITA)

A person having ordinary skill in the art as of the 2001-01-19 priority date would be a POS/self-checkout systems engineer with roughly a bachelor's degree in electrical engineering, computer science, or mechanical engineering plus 1–2 years of experience integrating bar code scanners, load-cell scales, and POS terminal software — or, equivalently, 3–4 years of practical experience in retail checkout design without a degree.

Two facts drive the obviousness conclusion:

1. The field was extraordinarily crowded and mature before 2001. The '791 patent itself cites ~52–57 references spanning 1972–2001, and its forward-citation list is 77 documents. A POSITA in January 2001 had decades of self-checkout art in front of them.
2. The claimed technical content is not hardware. The alleged point of novelty — hold data for multiple items in memory, then process it — is one of the most elementary constructs in computer programming. The patent discloses no special-purpose buffer structure; element 390 is simply part of main memory 370 (RAM).

---

3. Claim construction points that drive the obviousness result

---

4. Prefatory point: which references actually qualify as prior art

Before combining anything, several references printed on the '791 face are suspect as art against a 2001-01-19 priority date, because their publication dates postdate it:

• US 6,189,790 (NCR, issued 2001-02-20) — but its application was filed 1999-12-22, so it is available only as pre-AIA § 102(e) art, and only for what it discloses as of that filing.
• US 6,354,498 (NCR, issued 2002-03-12) and US 6,363,355 (NCR, issued 2002-03-26; filed 1998-12-21) — § 102(e) art at best.
• JP 2001184563 A (NCR, published 2001-07-06) — post-critical-date publication; not § 102(a)/(b) art, and a Japanese publication is not § 102(e) art under pre-AIA law. Treat as non-art absent more.
• US 2002/0096564 A1 (the '791 patent's own pre-grant publication) and US 7,044,370 (Bellis/ECR, provisional priority 2001-07-02) are not prior art to '791. Do not use them in a § 103 combination.

Conversely, the strongest reference on the face of the patent is unambiguously § 102(b) art: US 5,125,465 (Schneider), filed 1991-02-04, granted 1992-06-30 — more than eight years before the '791 critical date.

---

5. The keystone reference: US 5,125,465 (Schneider), "Fast retail security weighing system"

This reference is not merely "a" combination partner. On the verified text it discloses nearly the whole of the '791 concept, including the specific two-item and combined-weight scenarios the '791 patent treats as its invention.

Verified disclosures (source: https://patents.justia.com/patent/[5125465](/patent/5125465); https://www.freepatentsonline.com/5125465.html; PDF at https://patentimages.storage.googleapis.com/da/39/70/62004e22152db6/US5125465.pdf):

1. The same field and the same security problem. "Typically a product code is entered (manually or via a machine bar code reader) and then the purchased product is weighed and packed. The product code directly or indirectly (via a product lookup table database) indicates how much the product should weigh, and possibly, the allowable tolerances.... If the product's actual weight corresponds with the weight indicated by the product code, then the probability that the correct product has been weighed and packed is significantly greater than if no such correspondence is obtained." Schneider's background expressly discusses Ehrat, U.S. Pat. No. 3,836,755 and Mergenthaler, U.S. Pat. No. 4,779,706 — both of which also appear on the '791 face.

2. The buffer. Claim 1's typical embodiment recites "(e) a set of memory registers receiving said fourth signal for temporary storage" — i.e., expected-weight data for scanned product codes is held in temporary memory.

3. ★ Multiple items scanned and stored before verification of the first. "if a second purchased item's product code is entered before the weight corresponding to the first purchased item's product code has been measured and verified, the system will consider the weight on the scale as corresponding to the first purchased item alone, but if such weight does not correspond with the allowed weight for the first purchased product, then the system will consider whether the weight on the scale corresponds to the summation of the first and second purchased items' weights."

4. Combined-weight verification of two items. The controller "sums the contents of the said registers ... containing the weight information for the last purchased product with the registers ... containing the weight information for the second to last purchased product ... and verifies the said second controller signal against the said first controller signal."

5. Buffer clearing on verification. On a valid comparison, the controller "sends a first processor signal to said set of memory registers so as to erase the contents of the said memory registers containing the weight information for the last purchased product" (and, on the two-item path, for the second-to-last as well).

6. ★ The motivation to do exactly this. "A major limitation of the above inventions is that only one product can be weighed at a time.... The length of time required for an adequate representation of weight is generally 2 to 3 seconds.... in supermarkets where a human cashier is aided by a bagger, the cashier is generally able to scan 36 items per minute. If an automated point of sale system is to duplicate such productivity, then the automated point of sale system must be able to obtain an adequate representation of weight in effectively less than 2 seconds." Schneider's stated object: "to weigh a high rate of products per minute."

7. Also disclosed: a laser bar code scanner or keypad as the code input device (cl. 10–11 of '465); multiple alternative scales with an indicator telling the user which scale is free (cl. 13); operator prompting on verification failure (cl. 14); and a single-chip implementation (cl. 15).

Assessment. Schneider '465 discloses, or renders obvious, the following '791 claims on its own or with at most a trivial secondary reference: 1, 2, 3, 5, 7, 8, 9, 10, 11, 14, 15, 16, 18, 19, 23, 26, 27, 28, 32, 36, 37, 38, 39, 40 — subject to the two qualifications in § 10 below.

There is a genuine tension here worth stating plainly: Schneider '465 is marked as cited by the examiner on the '791 face, and the '791 specification's own background (and FIG. 5A) characterizes the prior art as requiring a strict "scan 500 → weight verify 510 → next item" sequence. On the verified Schneider text, that characterization is incomplete: Schneider expressly contemplates entering a second code before the first is verified and checking the summed weight. That gap between the admitted prior art and the cited prior art is the single most productive line of attack on this patent.

---

6. The other verified building blocks

I have not verified the full text of the many Tec, Dumont, Humble, Stambler, and NCR references in the list; they are used below only where their titles independently support a proposition, and I say so.

---

7. The combinations

Combination A — Schneider '465 + Checkrobot '018 (and optionally '343) ± Checkrobot '970

Targets: claims 1, 2, 3, 5, 7, 8, 9, 10, 11, 14, 15, 16, 18, 19, 23, 26, 27, 28, 32, 36–40.

What the combination supplies. Schneider supplies the multi-item buffer, the "second code before first verification" sequence, the combined-weight check, and the buffer-clear. Checkrobot '018 supplies (i) the explicit teaching that multiple articles may be simultaneously resident in the weighing zone and weighed at common times, (ii) deferral of the characteristic measurement to a location downstream of the code read (the security-tunnel paradigm — structurally the same "hold the identification, verify later" architecture), and (iii) the EAS layer. Checkrobot '970 supplies the alternative "size or weight or both" characteristic (claim 14's "one or more physical characteristics") and the operator override step that the '791 spec describes for the security-scale violation.

Motivation (KSR rationales):

• Same field, same problem, same solution mechanism. All three are operator-unattended retail checkout systems that read a UPC, fetch a stored characteristic correlated to that UPC, measure the article, and compare. There is no field-crossing or analogous-art argument available to the patent owner.
• Express design incentive and market force. Schneider states the throughput objective and quantifies it (36 items/minute; sub-2-second weighing). The '791 background admits the identical problem in nearly identical words ("bottlenecks at the checkout stations, reduced throughput, unhappy customers, and negative impact on the financial condition"). That is an admission of the motivation.
• Predictable result, no new structure. Every physical element of claim 1 — computer, memory, code reader, security mechanism — is old. The combination adds only the order in which already-obtained data is processed. Schneider had already built the exact registers and the exact two-item summation logic.
• Checkrobot explicitly hands over the multi-item teaching. "Accommodate plural articles ... and effecting the weighing scale at common times" is the multi-item verification the '791 spec presents as novel ("Allowing a customer to grab two items, one in each hand, and scan both items and place both items on the scale(s) at 290 at the same time ... greatly improves the speed").

Combination B — Any of the above + Symbol '485/'725 (statistical sampling) ± Symbol '735 (portable terminal)

Targets: claim 1's "multiple items" generality, claim 31 (N > 2), and the "variable amount" concept in the spec.

Motivation. Symbol teaches that the security check need not attach to every item, that items may be checked after a list has been compiled, and that the number of items checked may be varied by an algorithm (queue length, shopper history). Once a POSITA accepts that verification may be decoupled from the scan event and batched, the step to a buffer of arbitrary depth N is a routine parameterization. Symbol's "list of items self-scanned by the shopper is compiled and made available to a cashier" is a buffer of identification information for multiple items in everything but name.

Combination C — Any of the above + NCR '264 (post-scan item shuffling)

Targets: claims 21, 22, 24, 29, 34, 35 (verification in a different order than scanning).

Motivation. NCR '264 is directed to the exact problem created by decoupled scanning and bagging: because the customer "selectively choose[s] the order in which items are scanned" and "the order in which items are loaded into the grocery bags," the security system must "differentiate between entered items and unentered items." Its solution — detect a weight decrease on the post-scan shelf and a matching weight increase in the bag, and match them — is order-agnostic by construction. A POSITA who has adopted a multi-item buffer specifically to buy ordering freedom would consult the reference that addresses ordering freedom. This is the "use of a known technique to improve a similar device" rationale.

Combination D — Any of the above + Checkrobot '970 + NCR '447/'038-type camera art

Targets: claim 8 ("physical characteristic measuring device"), claim 14 ("one or more physical characteristics"), and the spec's camera/dimension embodiment.

Motivation. Checkrobot '970 states in terms that the measurable characteristic "can be article size, article weight, or both." Substituting a size/dimension sensor for a scale in a UPC-correlated-characteristic comparison is a substitution of known elements to obtain a predictable result — the security function (deterring price substitution) is identical.

Combination E — Single-reference attack on the two-item and combined-weight claims

Targets: claims 26, 32, 40 (and dependents 27, 28, 30, 33, 34).

Schneider '465, standing alone, discloses each element: scan a first item's code; scan a second item's code before the first is weighed and verified; hold both items' expected weights in temporary registers (the "same buffer"); place both items on the scale; obtain the combined weight; use the combined weight to verify identity of both items; then erase the registers. Claim 40 in particular maps almost element-for-element onto Schneider's fallback path.

---

8. Claim-by-claim vulnerability map

---

9. The patent owner's best rebuttals, and why they likely fail

1. "Schneider's registers hold only the last two items, not an arbitrary N."
   Likely to succeed only as to claim 31 (N > 2) and as a distinction over the "multiple items" generality of claim 1 if construed to require N ≥ 3. It fails as to claims 26, 32, and 40, which are expressly two-item claims. And it fails as an obviousness argument generally: extending a two-entry rolling buffer to an N-entry buffer is a routine parameter change with a predictable benefit (MPEP 2144.04 — "obvious to try"; changing a known parameter).

2. "Schneider's first check is for the first item alone; the summation check is only a fallback."
   Claim 40 requires only that the combined weight be "use[d] ... to verify identity of both the first item and the second item." Schneider does precisely that on the fallback path. Claim 1 requires only storing before verifying; ordering of the internal check steps is not claimed.

3. "Schneider is a weighing system, not a 'checkout system with a computer having memory with a buffer.'"
   Schneider's own claim 1 recites memory registers, a controller circuit, a product lookup table, and an output "whose destination is the host point of sale system." Claim 1's "computer" and "buffer" are not narrower than that. This is a labeling argument, not a technical one.

4. "The references are non-analogous / from different sub-fields."
   Unavailable. Schneider, Checkrobot, NCR, Symbol, and the '791 patent are all in the operator-unattended/self-checkout art, and they cite each other (the '791 face cites Schneider, Checkrobot '343/'018/'970, and NCR's '706; Schneider's own background cites '706 and '755, which also appear on the '791 face). A common citation network defeats any non-analogous-art theory.

5. "The prior art taught away — FIG. 5A shows it was thought necessary to verify each item."
   Weak. Teaching away requires that the reference criticize, discredit, or otherwise discourage the claimed solution. Schneider does the opposite: it criticizes the one-at-a-time requirement as "a major limitation" and teaches scanning a second item before verifying the first. A reference that denigrates the very sequencing the '791 background admits is the prior art is an explicit teaching, not a teaching away.

6. "The invention yielded unexpected results — universal industry adoption."
   No objective-indicia record exists: the patent was never asserted, never challenged at the PTAB, and expired 2021-01-19. Post-expiry industry-wide adoption of buffered self-checkout cannot carry a nexus to the claimed invention as a whole, particularly where the claimed combination produces exactly the predictable result Schneider promised (higher items-per-minute throughput).

---

10. Where the patent is comparatively stronger

Two honest qualifications, so this analysis is not read as a blanket invalidity conclusion I cannot support on the present record:

• Claim 31 (N > 2). Schneider teaches a two-deep register set. I have not verified a reference that expressly accumulates and later verifies an unbounded or user-configurable number of scanned items. Symbol '485 teaches a compiled self-scan list and a variable number of audited items, which I regard as sufficient for obviousness but which is not the same thing as a live N-item verification buffer. This claim is the patent's best § 103 position.
• Claim 20 (the "one in each hand" recitation). Purely human-behavior language; the machine-side capability is in Schneider, but a tribunal could give weight to the express handling recitation. I rate it moderate, not strong.
• Gaps in my evidence. I relied on verified full text for Schneider '465, Checkrobot '018 and '970, NCR '264, and Symbol '485. For the NCR self-service-terminal patents (5,747,784; 5,952,642; 6,032,128; 5,992,570; 6,056,087; 6,080,938; 6,155,486; 6,167,381), the Tec references (5,560,450; 5,590,614; 5,609,223; 5,662,190), Dumont, Humble, and Stambler I have only bibliographic data and titles. I have not built an element-by-element chart against their actual text, and none should be asserted against a claim without pulling the full documents.

---

11. Bottom line

The obviousness case against US 6,598,791 is strong as to the great majority of the claims, and it runs principally through a single reference the examiner already had before them.

• US 5,125,465 (Schneider) — a 1992 § 102(b) reference — discloses the retail-scale-plus-UPC security paradigm, a temporary register set holding expected data for scanned codes, the entry of a second product code before the first item's weight is measured and verified, a summation check of the first and second items' expected weights, and erasure of the registers upon successful verification, all motivated expressly by the need to beat one-at-a-time weighing and reach cashier-equivalent throughput. That maps onto independent claims 1, 10, 11, 14, 23, 26, 32, 36, 37, 38, and 40, and onto claims 2–9, 15–19, 27, 28, and 39.
• Combination A (Schneider + Checkrobot '018/'343 + Checkrobot '970) closes any residual gap on multi-item simultaneous weighing ("plural articles in residence ... weighing scale at common times"), the "size or weight or both" physical characteristic, and operator override — motivated by identical field, identical problem, and an express design incentive.
• Combination B (+ Symbol '485/'725/'735) supplies the compiled multi-item list and the variable/statistical basis for verification, addressing the "multiple items"/N-items generality.
• Combination C (+ NCR '264) supplies order-independent verification and the "entered vs. unentered" tracking that directly reads on the out-of-order claims.
• Diagnostic point for the patent owner's exposure: the '791 background characterizes the prior art as an inflexible scan-then-weigh loop, yet its own cited Schneider reference teaches breaking that loop. That is the cleanest way to frame the motivation to combine and the strongest rebuttal to any unexpected-results argument.

Practical caveat, consistent with the earlier sections: the patent expired 2021-01-19 and was never asserted or challenged. Any § 103 analysis is therefore aimed at (a) a pre-expiry damages window — which, on a 2026 complaint, the earlier section computed at under four months under § 286 — or (b) defensive/estoppel purposes, not at invalidating a live exclusionary right.

Confidence: High on the Schneider-based reasoning (verified primary text). Moderate-to-high on the combination rationales. Low on any claim-specific assertion resting on references I reviewed only by title. Before filing anything, pull the full texts of the NCR, Tec, Dumont, and Symbol '735 references and obtain the '791 file wrapper to determine why the examiner's Schneider citation did not result in a rejection.