Thanks.

Like I said, I could reproduce the issue after about 40 rows: ie, first 40ish rows were green, the other 60ish were not formatted at all. It may be that it's different on different machines (32 vs 64bit Win; 32 vs 64bit Excel; Excel version; Spreadsheet::WriteExcel version, etc¹). In this post, for example, Eily was able to "reproduce the defect with 300 lines of dummy data", so it may take a different number of rows for you or someone else.

Since I was able to replicate the problem with 100 lines of random data, I didn't put much thought into "well-conditioned". But yes, for SSCCE-purposes, the OP should have been able to post example data in <readmore>data</readmore> tags. But given that the newerlbie has declared that the issue is fixed, apparently the OP was on a system where it was hitting a similar limit to what Eily and I found.

And really, there's no good reason to have the $header1 format inside the loop, so putting it outside is arguably "better", even if you haven't hit the limit (ie, avoiding defining a large number of items which are all identical)

¹: My experiment was run with Spreadsheet::WriteExcel v2.40, Strawberry Perl v5.26.2 64bit, Win10.0.16299.611 64bit, Office 365 Excel 2016 MSO 16.0.9126.2259 32-bit