diff --git a/simap.h b/simap.h
index bdeff42..2955228 100644
--- a/simap.h
+++ b/simap.h
@@ -25,11 +25,11 @@ typedef union simap_ptr64 simap_ptr64;
  *
  * XXX: So beware that this can FAIL just "statistically most often works"!
  *
- * The memory layout after at *base is as follows:
+ * The per-elem memory layout after at *base is as follows:
  *
  * 8 byte:
  * - void* value;
- * - ? padding (only for non-64 bit pointer machines)
+ * - [?] optional padding (only for non-64 bit pointer machines)
  *
  * 8 byte:
  * - uint32_t previndex;
@@ -39,8 +39,6 @@ typedef union simap_ptr64 simap_ptr64;
  * - char name[]; // inline stored
  * - padding (divisible by 8)
  *
- * ELEMENTS added to it...
- *
  * Because of it a lookup is basically via strstr-like with 8byte steps!
  * with few character names zero-padded in the search term parameter
  * and if you want check extra validity by jumping back&forth in it.
@@ -50,7 +48,6 @@ struct simap_instance {
 	uint32_t end;
 	uint32_t prev_usage_end; /* previous usage_end or -1 if no previous exists... in bytes!!! */
 	uint32_t usage_end; /* in bytes!!! */
-	uint32_t next_previndex; /* in bytes!!! */
 	/** see doc comment for layout and why uint64_t* is the type */
 	uint64_t *base;
 };
@@ -62,7 +59,6 @@ static inline simap_instance simap_create() {
 	ret.end = 0;
 	ret.prev_usage_end = (uint32_t) -1;
 	ret.usage_end = 0;
-	ret.next_previndex = 0;
 	ret.base = ((uint64_t*) aralloc(&(ret.a), sizeof(uint64_t), sizeof(uint64_t), 1)) /* addr divisible by 8 */
 		+ 1; /* First really addressible thing */
 	return ret;
@@ -182,9 +178,8 @@ static inline void *simap_force_add_internal(simap_instance *map, const char *ke
 	data->ptr = ptr;
 
 	/* Save link to previous */
-	uint32_t *usprev = (uint32_t *)((uint8_t *)(map->base) +
-		sizeof(simap_ptr64) +
-		sizeof(uint32_t));
+	uint32_t *usprev = (uint32_t *)((uint8_t *)(map->base) + usi +
+		sizeof(simap_ptr64));
 	*usprev = previ;
 	/* and nex */
 	*(usprev + 1) = (uint32_t) -1;
@@ -193,8 +188,8 @@ static inline void *simap_force_add_internal(simap_instance *map, const char *ke
 	simap_c64 *start_str = (simap_c64 *)(usprev + 2);
 	*start_str = first8;
 
-	/* Remainin bytes */
-	if(keylen > 8) {
+	/* Remaining bytes */
+	if(keylen >= 8) {
 		/* uint32_t key_remains = keylen - 8; */
 		char *rem_str = (char *)(start_str + 1);
 		strcpy(rem_str, key + 8);