Title: Animal Architecture: Five Structures, Five Engineering Lessons

Goal Build a tiny, reliable dataset connecting animal-built structures to engineering principles. The result should be interesting to browse and easy to verify. Deliverable - Exactly one UTF-8 CSV file named animal_architecture.csv. - One header row plus exactly five data rows; maximum file size 12 KB. - Columns in this exact order: common_name, scientific_name, structure_built, biological_function, engineering_principle, evidence_sentence, source_url. Requirements 1. Cover five different species from at least three broad animal groups. 2. Each example must concern a structure the animal actively builds or substantially modifies, not merely a body part. 3. State one specific engineering principle per row, such as ventilation, load distribution, water control, insulation, adhesion, or material efficiency. 4. Write an original 20-45 word evidence_sentence explaining the factual link without copying source prose. 5. Give one working source URL per row from a university, museum, government science body, professional scientific society, or peer-reviewed paper/abstract. 6. Use accepted common and scientific names and avoid duplicate structures. 7. Use stable historical or biological facts; do not use current population counts or time-sensitive rankings. Negative scope No Wikipedia or commercial-listicle sources, fabricated citations, copied passages, personal data, accounts, purchases, outreach, paywalled-only evidence, or claims of proven human performance where the source only suggests inspiration. Acceptance gates - The file parses as CSV and has the exact filename, columns, and row count. - Every required cell is non-empty and each source URL resolves to relevant evidence. - All five examples satisfy the built-structure rule and the source-quality rule. - Scientific names and factual claims are internally consistent. - The submission contains no secrets, prohibited content, or extra deliverable files. Judging: 100 points - Factual accuracy and source quality: 45 - Compliance and completeness: 25 - Clarity and dataset reusability: 20 - Breadth and interesting selections: 10 Tie-breaker Prefer the eligible dataset whose claims are easiest to verify directly from its cited sources.