Models of average Galactic chemical abundances are in good general agreement with observations for [Fe/H] > –1.5, but there are gross discrepancies at lower metallicities. Only massive stars contribute to the chemical evolution of the ‘juvenile universe’ corresponding to [Fe/H] ≲ –1.5. If Type II supernovae (SNe II) are the only relevant sources, then the abundances in the interstellar medium of the juvenile epoch are simply the sum of different SN II contributions. Both low-mass (∼8–11 M⊙) and normal (∼12–25 M⊙) SNe II produce neutron stars, which have intense neutrino-driven winds in their nascent stages. These winds produce elements such as Sr, Y and Zr through charged-particle reactions (CPR). Such elements are often called the ‘light r-process elements’, but are considered here as products of CPR and not the r process. The observed absence of production of the low-A elements (Na through Zn including Fe) when the true r-process elements (Ba and above) are produced requires that only low-mass SNe II be the site if the r process occurs in SNe II. Normal SNe II produce the CPR elements in addition to the low-A elements. This results in a two-component model that is quantitatively successful in explaining the abundances of all elements relative to hydrogen for –3 ≲ [Fe/H] ≲ –1.5. This model explicitly predicts that [Sr/Fe] ≥ –0.32. Recent observations show that there are stars with [Sr/Fe] ≲ –2 and [Fe/H] < –3. This proves that the two-component model is not correct and that a third component is necessary to explain the observations. The production of CPR elements associated with the formation of neutron stars requires that the third component must be massive stars ending as black holes. It is concluded that stars of ∼25–50 M⊙ (possibly up to ∼100 M⊙) are the appropriate candidates. These produce hypernovae (HNe) that have very high Fe yields and are observed today. Stars of ∼140–260 M⊙ are completely disrupted upon explosion. However, they produce an abundance pattern greatly deficient in elements of odd atomic numbers, which is not observed, and therefore they are not considered as a source here. Using a Salpeter initial mass function, it is shown that HNe are a source of Fe that far outweighs normal SNe II, with the former and the latter contributing ∼24% and ∼9% of the solar Fe abundance, respectively. It follows that the usual assignment of ∼⅓ of the solar Fe abundance to normal SNe II is not correct. This leads to a simple three-component model including low-mass and normal SNe II and HNe, which gives a good description of essentially all the data for stars with [Fe/H] ≲ –1.5. We conclude that HNe are more important than normal SNe II in the chemical evolution of the low-A elements from Na through Zn (including Fe), in sharp distinction to earlier models.