diff --git a/pkg/vere/king.c b/pkg/vere/king.c
index 7446ae7199..74ff7f782d 100644
--- a/pkg/vere/king.c
+++ b/pkg/vere/king.c
@@ -251,6 +251,7 @@ _king_curl_bytes(c3_c* url_c, c3_w* len_w, c3_y** hun_y, c3_t veb_t)
   CURL    *cul_u;
   CURLcode res_i;
   long     cod_i;
+  c3_y     try_y = 0;
   uv_buf_t buf_u = uv_buf_init(c3_malloc(1), 0);
 
   if ( !(cul_u = curl_easy_init()) ) {
@@ -263,29 +264,34 @@ _king_curl_bytes(c3_c* url_c, c3_w* len_w, c3_y** hun_y, c3_t veb_t)
   curl_easy_setopt(cul_u, CURLOPT_WRITEFUNCTION, _king_curl_alloc);
   curl_easy_setopt(cul_u, CURLOPT_WRITEDATA, (void*)&buf_u);
 
-  res_i = curl_easy_perform(cul_u);
-  curl_easy_getinfo(cul_u, CURLINFO_RESPONSE_CODE, &cod_i);
+  while ( 5 > try_y ) {
+    sleep(try_y++);
+    res_i = curl_easy_perform(cul_u);
+    curl_easy_getinfo(cul_u, CURLINFO_RESPONSE_CODE, &cod_i);
 
-  //  XX retry?
-  //
-  if ( CURLE_OK != res_i ) {
-    if ( veb_t ) {
-      u3l_log("curl: failed %s: %s", url_c, curl_easy_strerror(res_i));
+    if ( CURLE_OK != res_i ) {
+      if ( veb_t ) {
+        u3l_log("curl: failed to fetch %s: %s",
+                url_c, curl_easy_strerror(res_i));
+      }
+      ret_i = -1;
     }
-    ret_i = -1;
-  }
-  if ( 300 <= cod_i ) {
-    if ( veb_t ) {
-      u3l_log("curl: error %s: HTTP %ld", url_c, cod_i);
+    else if ( 300 <= cod_i ) {
+      if ( veb_t ) {
+        u3l_log("curl: error fetching %s: HTTP %ld", url_c, cod_i);
+      }
+      ret_i = -2;
+    }
+    else {
+      *len_w = buf_u.len;
+      *hun_y = (c3_y*)buf_u.base;
+      ret_i = 0;
+      break;
     }
-    ret_i = -2;
   }
 
   curl_easy_cleanup(cul_u);
 
-  *len_w = buf_u.len;
-  *hun_y = (c3_y*)buf_u.base;
-
   return ret_i;
 }
 
@@ -376,6 +382,7 @@ u3_king_next(c3_c* pac_c, c3_c** out_c)
   }
 
   //  skip printfs on failed requests (/next is usually not present)
+  //REVIEW  new retry logic means this case will take longer. make retries optional?
   //
   if ( _king_curl_bytes(url_c, &len_w, &hun_y, 0) ) {
     c3_free(url_c);