The whole key s,q is not needed to determine t or u, so there are partial key dependences. Here, s>t, and q>u, so we can take t and u out of R. We must create R2 and R3 to match these two determinants, making the LHS of them keys and the RHS nonkeys. However, in R2 and R3 both LHS and RHS attributes have been made keys, and this is incorrect. Looking at R1, it states that s>q. There is no determinate which states this, so this option MUST be wrong. The whole key s,q is not needed to determine t or u, so there are partial key dependences. Here, s>t, and q>u, so we can take t and u out of R. This is actually true, as s,q>t and s,q>u also means s,q>t,u, which is as stated in R. However, it still is not normalised. You need to take into account the s>t and q>u determinants. This is actually a valid relation, as it is exactly what is stated in R. However, it still is not normalised. You need to take into account the s>t and q>u determinants. Consider the following functional dependencies

