Diagrams and the second homotopy group (Q819554)

An old conjecture of \textit{J. H. C. Whitehead} [Ann. Math. 42, 409--428 (1941; JFM 67.0738.04)] asks if \(K\) is a 2-dimensional connected \(CW\)-complex with non-zero \(\pi_2\), and \(L\) is formed by adding cells of dimension \(\leq 2\), then is \(\pi_2(L)\) nonzero? This is still unsolved. A more general problem asks to give a description of \(\pi_2(L)\) in terms of \(\pi_2(K)\). Suppose \(L+K\cup e^1\cup e^2\), \(t\) is the new generator determined by \(e^1\), and the attaching map for \(e^2\) represents a word \(w\) in \(\pi_1(K)*\langle t\rangle\). Under the two assumptions (1) \(\pi_1(K)\) is torsion-free and (more technical) (2) the \(t\)-shape of \(w\) is amenable, the authors succeed in proving that there is a \(\mathbb Z\pi_1(L)\) module isomorphism between \(\pi_2(L)\) and \(\mathbb Z\pi_1(L)\otimes_{\mathbb Z_{\pi_1(K)}}\pi_2(K)\). They also give interesting examples to show that the two hypotheses (1) and (2) are both necessary.