**The most surprising finding is that the same companies show up in several monthly files with amounts that don't reconcile. The "funding rounds" in this dataset can't be trusted at face value, and summing the Amount column would badly mislead.** I matched company names across the five monthly tables (SQL over the raw tables, 14 companies appearing more than once). The rendered table shows the pattern: - **Pine Labs appears in all four of January to April.** The amounts are $20M, $150M, $50M and $1.2B, and the April row's stage reads "undisclosed". Four successive rounds growing 60-fold in four months is implausible for one company. It looks more like cumulative or mis-keyed figures. - **Same company, same stage, very different amounts.** - DealShare is "Series E" in both February ($45M) and April ($624.8M). - Xpressbees is "Series F" in February ($300M) and April ($577.6M). - Licious is $150M in March and $488.7M in April. - Chargebee is $250M in February and $468.2M in April. - **Malformed amount strings, mostly in April.** Examples are `$624.800,000`, `$110.800,000`, `$129.500,000` and `$431.6`. The decimal point and comma are swapped (European-style formatting), or the value is truncated. Because Amount is stored as text, any naive numeric cast would silently drop or misread these rows, and they are the largest figures. - **Exact duplicates within a month.** NowPurchase appears twice in January with an identical $2.4M seed. Securden appears twice in April with $11.7M, once written `$11.700,000` and once `$11,700,000`, so a simple de-duplication on the raw text would miss it. - **Undisclosed amounts and blanks.** The most common Amount value is "Undisclosed" (35 rows), followed by 25 blank amounts. The next most common is $1,000,000 (20 rows). Together, "Undisclosed" and blank amounts outnumber any real dollar figure. - **The May file is mostly empty.** Of its 42 rows, 19 have no company name and nearly no other fields. Only one has an amount (`6400000`, with the stage mangled as `$Series A`), and it is in a different format from every other month. **Why it matters:** the dataset looks like a clean list of one funding round per company. In fact it mixes repeated or restated announcements, formatting errors and blank rows. Any total, average or ranking built from it without cleaning would be driven largely by the errors, especially Pine Labs' $1.2B and the April rows with swapped separators. **Caveat:** I can't tell from the data which figure is correct for each repeated company. The later amounts could be real follow-on rounds, restatements, or errors.