SS or TSS is the sum of the square deviation scores from the mean for one variable.
It's the error calculation of the no predictor model (The basis model).
<MATH> \begin{array}{rrll} \text{Total sum of squares (TSS|SS)} & = & \sum_{i=1}^{\href{sample_size}{N}}(\href{predictor}{X}_i-\href{mean}{\bar{X}})^2 & \text{for a predictor variable} \\ (TSS|SS) & = & \displaystyle \sum_{i=1}^{N}(\href{target}{Y}_i - \href{mean}{\bar{Y}})^2 & \text{for an outcome variable} \\ \end{array} </MATH>