Something similar goes for phrenology (the prediction of mental attributes from head shape): Science can discover where phrenology works and which traits it can predict and where it stops working. The sheer number of possible traits and populations one can correlate makes me confident you'd find something scientific if you looked. But if science discovers that elongated heads predict high extraversion among Swedes, and is otherwise largely wrong or unpredictive, would you say that phrenology is science?

Steven Kaas quipped something I find applicable: "Yes, I was wrong, but that only makes me falsifiable which makes me scientific which makes me right."This should be taken as general truth about the path towards becoming right - not a post hoc defense of something specific that didn't replicate.